Konstantinos Giamalis

Rule #5: Single source of truth for every entity

Sun, 13 Sep 2026 13:00:00 GMT

Fifth rule.

Every entity lives in exactly one place.

A customer. A building. A contract. A request. A supplier. Each one has one home, and everything else points to it. Not a copy. A pointer.

The moment the same fact lives in two places, one of them is wrong. You just don’t know which yet. The phone number in the CRM says one thing, the spreadsheet says another, the email signature a third. Three people act on three truths and all of them are doing their job.

A man with one watch knows the time. A man with two is never sure.

This is not a software problem. It starts with a decision nobody wants to make: which system is the record for this entity? Say it out loud. Write it down. From now on, the contract lives here. If it’s not here, it isn’t signed. If the address changed, it changed here first.

Everything else becomes a view. The dashboard reads from the source. The report reads from the source. The chatbot reads from the source. Views can be wrong for a minute. The source can’t.

The word to be suspicious of is sync. When two systems need to be kept in sync, you’ve already lost. You have two sources, and a job that fails on weekends. Better to have one source and many readers.

It gets harder as you grow. Every new tool arrives with its own idea of what a customer is, and it wants to own it. Say no. The tool can read. It can’t own.

The payoff is quiet. Arguments about “which number is right” stop, because there’s only one number. When a fact is wrong, you fix it once. And every process built on top, every report and every automation, inherits the correction without knowing it happened.

One entity, one home. Everything else is a window into it.

Rule #1: Do not lose information

Sun, 13 Sep 2026 12:00:00 GMT

I’m rebuilding the operations of a company from scratch. Not the product. The machine behind it: how work comes in, how it moves, how it gets done.

I’ve learned a few rules the hard way. I’ll write them down one at a time. This is the first.

Do not lose information.

Most operational chaos is not a people problem. It’s a plumbing problem. Information leaks. A request comes in by phone and lives in someone’s memory. A change is agreed in a chat and never reaches the system. A document is sent by email and dies in an inbox.

Every one of those leaks becomes a mistake three weeks later. Nobody can trace it, so everyone blames everyone.

Think of a hospital. If a patient’s allergy is written on a sticky note at reception, the surgeon doesn’t have it. Nobody was careless. The system simply allowed information to exist where it couldn’t be used.

So the rule is strict. Information enters through specific doors, and only those doors. Anything that didn’t come through a door doesn’t count. It’s not information. It’s rumour.

That sounds harsh, and it is. The first weeks are painful. People will say “but I told you”. And the answer has to be: not through the door, not in the system.

It’s also the single most powerful thing you can do. When you have all the information in one place, half of the problems you thought were process problems simply disappear. They were never process problems. They were missing facts.

Close the leaks first. Everything else is built on top.

Rule #2: Map all processes

Sun, 13 Sep 2026 11:00:00 GMT

Second rule for building an operations machine.

Map everything.

A process that isn’t mapped doesn’t exist. Not for the system. It exists in somebody’s head, and heads go on holiday, get sick, quit.

I resisted this for a long time. Mapping felt bureaucratic. Flowcharts. Documents nobody reads. The kind of thing large companies do instead of working.

Then I watched what happens without it. Every unmapped process is handled “the way we always do it”, which means five different ways by five different people, none of them written down. When something goes wrong, there is no way to find out which step failed, because nobody agreed on the steps.

Think of a kitchen. A good restaurant doesn’t have talented cooks improvising. It has a line. Every dish has a sequence, every station knows its part, and a new cook can plate the same dish on day one. Talent goes into the recipe once. Then the line repeats it a thousand times.

Mapping is not documentation. It’s the act of deciding, once, how something is done. The document is just the receipt.

And it has a brutal side effect. When you try to map a process, you often discover it isn’t one. It’s three exceptions and a habit. That’s the moment you find out what you’re actually running.

Map all of it. The boring ones, the rare ones, the ones only one person knows. Especially those.

If it isn’t on the map, the system can’t run it. Someone will, in their own way, until they don’t.

Rule #3: Add constraints

Sun, 13 Sep 2026 10:00:00 GMT

Third rule.

Harden the system. Don’t allow everything.

Flexibility sounds like a virtue. Sometimes it’s even dressed up as a customer-centric approach: we’ll do whatever the client asks, however they ask it. It feels generous. It doesn’t scale. In operations, every option you leave open is a place where a mistake can happen. Every free-text field is a future cleanup. Every “we’ll handle it case by case” is a case nobody will handle.

Think of a train. It can only go where the rails go. That’s not a limitation. That’s why it can carry a thousand people at 300 km/h and arrive on time. A car can go anywhere, and that’s exactly why it crashes.

Constraints are how you buy reliability. A required field. A status that can only move forward. A request that can’t be created without an owner. A form instead of a phone call. Each one removes a category of error before it happens.

People push back. Constraints feel like distrust. They’re the opposite. A constraint means I trust the system enough to not need heroics from you every day.

The trick is where you put them. Constrain the inputs and the transitions, not the people. Nobody should have to remember the rule. The system should make the wrong thing impossible, or at least annoying.

Robust doesn’t mean rigid. It means fewer ways to break. A system that allows everything handles nothing. And the customer you were trying to please is the first one it fails.

Rule #4: Structured data

Sun, 13 Sep 2026 09:00:00 GMT

Fourth rule.

Structure your data.

Once you stop losing information and you constrain how it comes in, something has quietly happened. Your data arrives in a shape you decided. Not prose. Not screenshots. Fields, states, dates, amounts.

That’s not tidiness. That’s the difference between a workshop and a factory.

In a workshop, a skilled person looks at each job and figures it out. It works, and it doesn’t scale. In a factory, parts arrive in a known shape, so the same step can run a thousand times without a decision. The intelligence moved from the person to the design.

Structured data is what lets a process be repeated. It’s also what lets it be verified. You can check a field. You can’t check a paragraph.

And this is where it connects to the tools we all now have. Andrej Karpathy made the point that LLMs do their best work where results are verifiable. Clean inputs, checkable outputs. The vaguer the input, the more the model guesses, and guessing at scale is just automated error.

Structure isn’t tidiness. It’s the design of the factory floor. Get it right and the same step runs a thousand times without a decision. Get it wrong and every task is a workshop again, with a smarter tool.

Switching from Super to Astro

Sun, 13 Sep 2026 00:00:00 GMT

This blog moved last night. From Notion and Super to Astro, a static site in a git repo, hosted on Cloudflare.

Kudos to Dimosthenis Avgeris, who suggested the switch.

Why? Faster. Cheaper. And the excuse to test a new tool. It always matters to check the good tools, even when the old one works.

How? I asked Fable 5.1, in ultracode mode, one question: is this a good idea?

Short answer: no.

It ran a full analysis. Sources, verification, a decision memo. Stay where you are, it said. The gains don’t justify the hours.

Ok. Do it.

It took less than an hour. Export, conversion, design, redirects, DNS. Not a single sophisticated prompt. Just plain sentences, one after the other.

Two things I keep from this.

Thanks, Dimo. A good nudge from someone you trust is worth more than a good analysis.

And reasoning still matters. The model was right on every fact and wrong on the decision, because it measured hours and dollars and I wanted something else: to own my words and to publish the way I write code now. A file, a commit, done. The machine gives you the answer to the question you asked. Asking the right question is still your job.

On mistakes

Sat, 27 Jun 2026 00:00:00 GMT

Progress is made of mistakes.

Not despite them. Made of them.

Every step forward carries a few wrong ones inside it. You don’t get the progress without the errors. That isn’t a motivational line — it’s just the arithmetic of moving fast with incomplete information.

So the job was never to avoid mistakes. The job is to handle them.

And handling them gets harder, not easier, the more experienced you become.

Early on, mistakes are cheap. Small blast radius. Nobody’s watching. You make one, you fix it, you move on.

Later it changes. The decisions are bigger. More people depend on them. And your own experience starts working against you — you’ve been right enough times to trust yourself more than you should. Ego and stakes rise together. The mistakes get more expensive exactly when they get harder to admit.

That’s the trap of experience. It doesn’t stop you from being wrong. It makes being wrong cost more.

The first discipline is to keep most mistakes reversible.

Some doors only open one way — the Type 1 decisions. The bet you can’t unwind. The hire you build the whole org around. The promise you can’t take back. You don’t get to be casual there.

Everything else, you should be casual about. Decide fast, stay cheap, keep the option to walk it back. Spend your caution on the one-way doors and almost nowhere else. Most mistakes are survivable. A few are not. Knowing the difference is most of the game.

And then there’s the part nobody likes.

A mistake isn’t fatal because it’s big. It’s fatal because it stays hidden.

The reversible mistake you catch early stays reversible. The same mistake, six months buried, becomes a one-way door — not because it changed, but because you ran out of room to undo it.

So the thing you actually optimize for isn’t fewer mistakes. It’s faster detection. Short feedback loops. People who feel safe telling you you’re wrong. A culture where bad news travels fast, instead of getting polished before it reaches you.

Companies rarely die from one big error. They die from a hundred small ones nobody surfaced in time.

And you don’t only have to learn from your own.

Other people’s mistakes are cheaper than yours — you get the lesson without paying for it. So read biographies. They’re full of expensive errors someone already survived and wrote down, so you don’t have to repeat them.

And talk about your own mistakes out loud. The moment you share one, people tend to share theirs back. That trade is one of the best deals there is — you walk away with a dozen failures you never had to live through.

So I’m not trying to make fewer mistakes anymore.

I’m trying to make them small, make them reversible, and make them loud.

On writing more

Mon, 25 May 2026 00:00:00 GMT

Writing a blog post used to be a project.

Here’s what it actually took. Think of the idea. Write a draft. Edit it. Hunt down the right links to my other posts. Set it up to go live — properties, date, the works. Find an image. Add the right tags.

Seven steps. And here’s the part I never admitted to myself: only the first one was writing.

The other six were logistics. They had nothing to do with the idea. They were a tax I paid every single time — and the tax was high enough that a half-formed thought rarely survived to become a post. The friction did the deciding for me. If an idea wasn’t worth an evening of formatting and link-hunting, it quietly died.

That was the real cost. Not the hours — the posts I never wrote because publishing was a chore.

Here’s what a post takes now:

I share my idea, the outline, and what I’m trying to say.
I check what Claude wrote.
Done.

The draft, the edit, the links, the live setup, the image, the tags — Claude handles all of it. This post went from a thought in my head to a published page in about the time it took to drink a coffee.

I want to be precise about what changed, because it’s easy to get this wrong. AI didn’t make me a better writer. It didn’t have the idea. It can’t tell me whether a thought is worth your time — that part is still mine, and it should be.

What it removed was the tax. The six steps that were never the point.

And that changes something bigger than my evening. Every chemical reaction has an activation energy — the minimum push needed before it happens at all. Publishing, for me, had a high activation energy. Most of my ideas never cleared it.

Now it’s close to zero.

So here’s my honest prediction: I’m going to write a lot more. Not because I suddenly have more to say — I always had more to say. It’s that the ideas which used to die in the gap between “interesting thought” and “worth seven steps” finally have somewhere to go.

This blog is about to get busier. That’s not a New Year’s resolution. It’s just what happens when you remove the thing that was rationing the output.

Relevant posts:

The tactical trap

Mon, 25 May 2026 00:00:00 GMT

I’m deep in the tactical right now.

Solving problems on the spot. Sitting close to operations. Feeling the pulse of the service every single day. I can see exactly what we need to build to make it better.

It feels great. And that’s the problem.

Tactical work is not the enemy of strategy. It’s a trap precisely because it works. It gives you an immediate result. It makes you feel productive. And it never runs out.

This isn’t procrastination. Procrastination feels bad — you know you’re avoiding something. Tactical work feels like the opposite. It feels like the most responsible thing you could possibly be doing right now. That’s what makes it so hard to catch.

Part of the pull is external. The fires are infinite. Operations will never hand you a quiet week.

But part of it is internal, and that part is harder to admit. It’s the builder mentality. You became a founder because you like building. Fixing. Shipping. Touching the product. Being in the tactical scratches that itch perfectly.

That instinct is an asset — it’s the reason you can build anything at all. But left alone, it quietly makes the same choice for you every morning: tactical over strategic. You don’t decide to skip strategy. You just never arrive at it.

And the cost is invisible until it’s expensive. You optimize the service beautifully — but you might be optimizing your way up the wrong hill. A company that is all tactical is very busy going somewhere it never chose.

I don’t think the answer is “do less tactical.” Being close to operations is real signal. It’s how I know what to build. The answer is to change my relationship with it.

Two things I’m starting to test.

The first: make the tactical an input, not just an output. Right now every problem I solve produces a fix and then disappears. So I’ve started keeping a log — what I put out today, and what it tells me. Done consistently, strategy time stops being a blank page. It becomes pattern-matching across everything the tactical already taught me. The tactical starts feeding the strategy instead of replacing it.

The second: delegate the tactical I’ve already learned from. I’m in the weeds partly to understand operations — and that’s correct. But once I’ve extracted the signal from a category of problems, that category no longer needs me. The first ten times I handle something, I’m learning. After that, I’m just a bottleneck. Delegation isn’t giving away all the tactical. It’s giving away the part I’ve already mined.

So here’s where I’ve landed, at least for now.

Keep the tactical. Stay close to the pulse. But make every piece of it earn its place. Either it’s still teaching you something — or it has already taught you, and it belongs to someone else.

The builder mentality that pulls you in is worth keeping. It just can’t be the one holding the steering wheel.

I’m testing this now. We’ll see.

Relevant posts:

Leadership lessons from Kolokotronis

Fri, 13 Feb 2026 00:00:00 GMT

Theodoros Kolokotronis was the military leader of the Greek Revolution of 1821. He participated in almost every important battle, and his strategic mind was the reason the revolution survived its most critical moments. But what truly separated him from every other leader of the war wasn’t tactics. It was leadership.

I tried to identify his most important leadership attributes, so I read a couple of books about him, plus his biography. Here’s what stood out.

Experience

Let’s start with basics. He was not an occasional military leader. Theodoros was fighting the Ottomans from a very young age. His father was killed in battle when he was just a boy — a battle where he and his mother barely survived. He fled the Peloponnese after years of being hunted, and made his way to Zakynthos where he enrolled in the British army under Richard Church. There, he received formal military education and experience in tactical warfare.

The combination of European military training with the guerrilla tactics he had practiced his entire life made him uniquely dangerous. Add to that his intimate knowledge of every path, cave, bottleneck and terrain advantage across the Peloponnese, and you had a leader who could read the battlefield like no one else.

Experience is not years. Experience is exposure to different environments, combined with the ability to synthesize.

Innovation

In the battle of Grana, the mission was clear: don’t let the Ottomans return to Tripoli with their reinforcements and supplies. The Greeks positioned themselves between the city and the retreat route. When the first wave of enemies arrived, the battle started.

But then, Ottoman forces from Tripoli came to support their side. The Greeks were suddenly caught in between two enemy fronts. A disaster in the making.

Kolokotronis improvised. He ordered his men to fight “κώλο με κώλο” — back to back — so they could engage both sides simultaneously. This two-sided fire formation (“δίζυγο πυρ”) had never been deployed as a tactic before. It was the right decision, at the right time, invented under pressure.

Innovation doesn’t come from brainstorming sessions. It comes from pressure.

Emotional Intelligence

After the Battle of Valtetsi, his army was marching along a path when they encountered dead bodies of Greek soldiers. The sight horrified his men — if these soldiers died, they could be next. Morale was collapsing in real time.

Kolokotronis walked up to the dead, kissed their remains, and told his army: these men became holy. They died for the sacred purpose of freedom. In an instant, he reframed death from a source of terror into a source of honor. His men found the courage to keep fighting.

On a different occasion, they visited the house of an Ottoman general, Kiamil Bey. His brother walked in, saw a dog sitting on an expensive carpet, and kicked it — not because he hated dogs, but because the luxury disgusted him. Greeks outside were struggling to survive, and here was a palace dripping with wealth. The tension was escalating fast.

Kolokotronis defused it instantly. He turned to his brother and said: “Come on, just because a dog bit you yesterday doesn’t mean all dogs are bad.” A simple reframe. Conflict avoided.

He was also imprisoned during the Greek civil wars — arrested by his own side. When Ibrahim Pasha invaded and the revolution was on the verge of collapse, they released him and handed him command of the entire Peloponnese army. He came back without bitterness, without settling scores. He focused on the mission.

That’s not just intelligence. That’s emotional discipline.

Decisiveness

When Ibrahim Pasha landed in the Peloponnese in 1825, he didn’t just use military force. He used a psychological weapon: the “proskynima” — a systematic campaign to make Greek villages individually surrender and submit to Ottoman rule. One by one, villages were folding. The revolution was dying from within.

Kolokotronis understood that this wasn’t a military problem. It was a willpower problem. And he responded with absolute decisiveness.

“Φωτιά και τσεκούρι στους προσκυνημένους.” Fire and axe to those who surrender.

He made the Greeks fear him more than they feared Ibrahim. Any village that refused to return to the Greek side was attacked. The leaders of the surrender movement were hunted down. It was brutal. It was necessary. The revolution was literally dying, and half-measures would have ended it.

The moment the most prominent collaborator, Nenekos, was executed, the tide turned. Villages came back. The revolution survived.

Sometimes, the hardest leadership decisions are not about strategy. They’re about drawing a line and enforcing it when everyone around you is ready to quit.

Strategic Thinking

The Dramali invasion is a masterclass in strategy.

In 1822, Mahmud Dramali Pasha marched south with an army of 30,000 — the largest force seen in Greece in over a century. The Greeks were in disarray. Government officials were fleeing. It looked like the end.

Kolokotronis didn’t panic. He didn’t try to fight a 30,000-strong army head-on with 2,500 fighters. Instead, he executed a multi-layered plan.

First, scorched earth. He burned crops, poisoned wells, destroyed food supplies across the Argolic plain. Second, he used Larissa castle as a decoy — Ypsilantis held it just long enough to buy time, even though it had no water supply. Third, he ordered every Greek soldier to light multiple campfires at night, creating the illusion of an army far larger than what actually existed. The Ottomans saw fires across every hill and believed they were surrounded by thousands.

Dramali’s army, starving and demoralized in the summer heat, was forced to retreat through the narrow Dervenakia pass. That’s where Kolokotronis was waiting. The result was annihilation. The revolution survived its greatest test.

He didn’t win because he had more soldiers. He won because he controlled the variables that mattered: terrain, supply, time, and psychology.

Faith

Across all these moments — the battles, the betrayals, the imprisonment, the civil wars — one thing remained constant. Kolokotronis never lost faith in the cause.

When Ibrahim was burning the Peloponnese and villages were surrendering, the revolution looked finished. But Kolokotronis held. He used guerrilla tactics, avoided open battle, and patiently wore Ibrahim’s forces down, waiting for the European powers to intervene. He couldn’t stop all the destruction, but he kept the flame alive.

He was patient when patience was the hardest choice. He endured when endurance seemed pointless.

That’s the lesson that’s hardest to teach and hardest to learn. Strategy can be studied. Tactics can be practiced. But the ability to keep going when everything says stop — that comes from something deeper.

Kolokotronis had it. And that’s why Greece exists today.

On constraints

Fri, 13 Feb 2026 00:00:00 GMT

We’re building a company with limited budget, limited time, and funding that’s enough to move but not enough to waste. Every euro counts. Every sprint counts.

That’s not a complaint. That’s the best thing that could have happened to us.

Constraints force decisions. And decisions force clarity.

When you have unlimited budget, you “explore.” You build features nobody asked for. You hire ahead of need. You overcomplicate things because you can afford to. When you have constraints, you build what matters. You cut the noise. You ship.

Here’s a real example. We needed a task management system. The brief was wide open — technically, we could build anything. Custom workflows, automations, integrations, the works. A system that “does everything.”

And that’s exactly where the danger was.

A system that does everything does nothing well. We’ve all seen it. The internal tool that became a monster. The Notion workspace with 47 databases that nobody understands. The “flexible” platform that requires a PhD to configure.

So we started with constraints. On purpose. What’s the minimum this needs to do? What are we NOT building? What’s the simplest version that still solves the problem?

That changed everything. The constraints didn’t limit us — they guided us.

There’s actually solid theory behind this. Eliyahu Goldratt introduced the Theory of Constraints back in the ’80s, originally in a book called The Goal. The core idea is simple: every system has a bottleneck, and the performance of the entire system is determined by that single constraint. Instead of trying to improve everything at once, you identify the constraint, exploit it, and subordinate everything else to it.

Goldratt’s five steps: identify the constraint, decide how to exploit it, subordinate everything else, elevate the constraint, and repeat. It was written for manufacturing, but it applies perfectly to startups. You don’t have infinite resources, so find the one thing that’s actually blocking progress and fix that. Ignore the rest.

Most founders I talk to think constraints are a problem to solve. Something to overcome once the next round closes. I think the opposite. Constraints are the reason you’re building something focused instead of something bloated.

The companies that scare me are not the well-funded ones. It’s the small, constrained teams that ship fast and stay focused. They don’t have the luxury of indecision.

Embrace your constraints. They’re not holding you back. They’re keeping you honest.

Behind Closed Doors

Thu, 12 Feb 2026 00:00:00 GMT

The best knowledge I’ve ever received didn’t come from a LinkedIn post. It didn’t come from a thread. It didn’t come from an influencer with 200K followers telling me how to “10x my productivity.”

It came from discussions behind a closed door.

A dinner with someone who’s been through hell building a company. A private call where someone shared exactly how they closed a deal — the ugly parts included. A WhatsApp message at midnight from a founder who said “don’t do what I did.”

That kind of knowledge is not politically correct. It’s not packaged for engagement. It doesn’t have a catchy hook or a carousel format. It’s raw, unfiltered, and it comes from people who have actual skin in the game.

I’ll make one exception here: Ratchisky. Exceptional content creator. But he’s the exception, not the rule.

Here’s something I noticed about my own journey. When I was an executive, I used to share a lot. Lessons learned, frameworks, strategies — you name it, I posted it. I even wrote a book of what actually worked for the Marketing team to scale and succeed. I was open about what worked and what didn’t. I genuinely enjoyed it.

Now, as a founder, I share almost nothing.

Not because I became a different person. But because I realized that the truly valuable insights — the ones that give you an edge — stay within the inner circle. They have to. The market is competitive, the stakes are higher, and the knowledge that actually moves the needle is not the kind you broadcast.

The real playbook is never public.

Think about the people you admire most in business. Not the ones with the biggest audience. The ones with the biggest results. Most of them are in silent mode. They don’t post. They don’t engage in debates. They build. They ship. They let the product do the talking.

Kudos to every builder out there operating in silence. The ones who wake up, do the work, solve the hard problems, and let their service and product speak for them.

The loudest room is rarely the smartest one.

Signal vs Noise

Fri, 06 Feb 2026 00:00:00 GMT

I knew running a startup would be hard. Everyone knows that.

What I didn’t know is how different it feels when you’re actually inside it.

Being a co-founder is not the same as being a C-level executive. I understood that intellectually. I didn’t understand it emotionally.

I’ll keep writing on this blog, but from here on these thoughts come from the co-founder seat. Less theory. More friction.

The thing that surprised me the most so far is the noise.

There is so much of it.

And the worst part is that most of it makes sense.

Product improvements that should happen. Marketing ideas that sound smart. Operational fixes that would clearly make things better. Hiring discussions. Small inefficiencies that feel bigger than they probably are.

None of these are wrong. That’s what makes it exhausting.

Before, as a CPO, prioritization was hard but contained. The fundamentals were already there. The company knew what it was. The direction was set. I was deciding what mattered more, not what mattered at all.

Now it’s different.

There is no clear baseline. You’re not choosing between good and bad. You’re choosing between many good things and doing almost none of them. And you don’t get to know if you were right until much later—sometimes too late.

Another thing I didn’t expect is how much this is about losing.

You can’t win most battles. You have to decide which ones you cannot afford to lose, and consciously let the rest go. Ideas you like. Decisions you believe in. Paths you won’t explore.

And you carry those losses with you. Quietly. Every day.

Frameworks don’t really help here. Not in the way we like to pretend they do. Things move too fast, information is incomplete, and waiting for clarity is usually the wrong move.

So you rely on instinct. On customer conversations. On a vague but growing sense of where the business wants to go. You pick a direction and commit, knowing it might be wrong.

You’re not filtering noise into a clean signal.

You’re choosing one signal among many and hoping it’s the right one—for now.

This is what the work actually feels like.

And I’m still learning how to sit with it.

On delegation

Thu, 19 Jun 2025 00:00:00 GMT

I had a chat yesterday with my friend Chris I. about how we learned to delegate tasks to our team.

I still remember a session with Yannis M., the most experienced Agile coach in Greece, seven years ago (!) when we were about to scale the Product team at efood.

Yannis said, “We need to significantly scale the team.”

I asked, “Why? I just need two PMs to help with XYZ tasks.”

He replied, “Your current model won’t scale. You handle too much on your own.”

I wondered, “Am I handling tasks the wrong way?”

He said, “No, but that approach doesn’t work. You must build a team you trust and delegate well. That lets you focus on the strategic side of your role. It is less tangible than completing tasks, but it matters and you’re a good fit for it.”

At first, I found it hard to let go of tasks, especially those in my core skill area, product marketing. Over the years, I improved. But I never paused to think how I learned. Yesterday’s chat made me realize it was mostly by doing and by failing.

No courses. No books. No formal trainings with real value (although I had multiple ones). Just action and feedback from colleagues. I tried frameworks like the Eisenhower Matrix, but I never made them work in real life.

In order to become better at delegation, I needed:

Humility to accept that, since I didn’t know how to delegate, I would make many mistakes.
Business understanding to avoid risking important projects: start with simple ones, but gradually delegate more critical tasks when the odds of success are high.
A strong team, which requires good hiring skills. I was lucky to build a great team, and that made delegation much more effective.
Patience, because it won’t work perfectly from day one.
No ego. It’s not about you anymore; it’s about your team. You succeed through your team, not through yourself. It may sound cliché, but it’s the most important lesson. Not all individual contributors are meant to be managers, and that’s okay.

PS: I truly believe a good manager should have been an IC first, but I might be biased since that was my path. I’ll write another post on that topic specifically.

Eisenhower matrix insight: Delegate the not important but urgent tasks 🤷‍♀️

Best strategy tip with 2 letters

Tue, 27 May 2025 00:00:00 GMT

Business strategy. Product strategy. Marketing strategy. A business should have a strategy that cascades and align with the strategies of all departments. At least in theory.

I read a couple of books (Good strategy / Bad strategy is a good one), I’ve been to multiple strategy sessions, from big companies (DeliveryHero), to high growth startups (instacar).

Generally, it’s a simple process, there are some basic principles in place but its mostly custom to fit the needs of the company and the participated audience.

In every strategy session I always try to remind myself and the audience the following:

In which projects will we say NO?

There are high-priority projects, there are critical projects, there are innovative projects but what truly matters is to define which projects are out of scope. We will not do them.

Saying no has multiple advantages:

It’s clear priority assessment.
It helps you focus.
It keeps you accountable.

The best strategy sessions I’ve been, the NO projects were clearly defined. Try it, it’s not easy but quite effective.

How Many Project Management Tools?

Sun, 11 May 2025 00:00:00 GMT

We live in the age of LLMs, AI agents, and nonstop talk about “optimized efficiency.” LinkedIn is packed with posts about running entire businesses on AI alone. But real life is harder than it looks.

Lately, I’ve spoken with dozens of professionals across Greece and Europe. A sad pattern keeps popping up: many companies are juggling four, five, even more project management tools. Asana for customer care. Linear for developers. JIRA for support tickets. Trello for roadmaps. Basecamp for daily check-ins. The list never ends. It’s crazy.

You might think it’s fine for each team to pick its favorite app. But this creates chaos. Here’s why one tool is better for everyone:

A single source of truth.
Better planning.
Clear communication.
Unified way of working.
Shared priorities.

Stop letting each team pick its favorite software. Pick one. Train everyone on it. Stick with it. You’ll save time, cut frustration, and get real visibility into what’s happening across your whole company.

One tool. One view. One way forward.

Scope vs Deadline

Mon, 14 Apr 2025 00:00:00 GMT

A couple of days ago, I watched a presentation by a friend of mine (hello, George Hadjivarnava) about a technical project’s scope and deadline.

Project scope typically includes:

User flow & designs
Project features
Technical specifications
Tracking
Project milestones, etc.

Deadline typically includes:

Deadlines per milestone
Final go-live deadline

These two concepts often conflict. Let me illustrate with a short matrix:

Scope	Deadline	Outcome
Large	Tight	Half the scope gets dropped
Large	Large	Scope evolves due to its size, and the deadline gets extended
Tight	Tight	The product will likely launch as an MVP and require further development
Tight	Large	This scenario does not exist

💡

What should we prioritize, scope or deadline?

The answer depends on the project and its context. But generally, when I have to choose, I go with the deadline.

Deadline is real constraint.

Deadline connects the deliverable with the reality.

Deadline help you become better in prioritization.

Deadline kills procrastination.

Here is the presentation made from George:

Open PDF

One important disclaimer: when pushing for deadlines, you must be very careful and sensitive with the scope. Otherwise, there’s a real risk of launching a product that falls short of customer expectations—and that can backfire.

Simple & Easy?

Wed, 09 Apr 2025 00:00:00 GMT

In the last few days, I’ve had conversations with several PMs who are about to launch new businesses, side hustles, or passion projects.

Right now, because of the AI revolution (hello MCPs), a truly good PM can (literally) build and run a business effectively. We’re seeing more and more tech CEOs who started as PMs (hello Sundar).

But this post isn’t about PMs turning into CEOs.

What struck me during these discussions was that, regardless of the stage—startup, scale-up, or corporate—the principles we talked about were always the same. Simple. Fundamental.

“Build the right UX for your customers”
“Offer customer support that responds in minutes and solves most problems”
“Create a lovable brand”
“Spend your marketing budget wisely”
“Innovate to keep your customers happy”

Sound like best practices? They are.

Sound cliché? Also true.

Sound like everyone is doing them? Yep.

Will you see them in every business deck, pitch, or vision board? Absolutely.

But here’s the paradox:

Although everyone preaches these fundamentals, only a few companies in the world actually do them well.

Think about it:

How many truly lovable brands can you name?
How many companies have above average customer support?

So why aren’t these simple, obvious things consistently applied?

Because they’re not easy.

Some are incredibly difficult.

Some are just very difficult.

None of them are easy.

All of them are simple.

💡

One of the smartest guy ever existed (Leonardo Da Vinci) once said:

Simplicity is the ultimate sophistication.

Here is a relevant TEDxTalk:

[Notes] PM Habits Seminar: Get Shit Done by Shreyas Doshi

Fri, 23 Aug 2024 00:00:00 GMT

I recently watched Shreyas Doshi’s webinar, “PM Habits,” which offers practical tactics that you can start implementing immediately. While some of these strategies may seem obvious or might not perfectly fit your situation, the key is to grasp the underlying principles and tailor them to your needs. Here are the notes I kept, segmented into topics:

The Power of Small Changes

One of the fundamental lessons from the webinar is the immense value of small, consistent changes. While large-scale transformations are often celebrated, it’s the incremental improvements that truly compound over time. As a product manager, embracing this mindset can lead to substantial long-term benefits.

Preparation is Key

Preparation lies at the heart of effective product management. To enhance your efficiency, Shreyas emphasizes making habits easy to adopt and rewarding to maintain.

Mastering Meetings

Product managers often find themselves in various types of meetings, each with its own set of challenges:

Imposed Meetings: These are recurring meetings that you must attend.
Influenced Meetings: Meetings where you have some say in the agenda.
Initiated Meetings: Those you schedule and lead.

To make the most of these meetings, always share the agenda upfront. This simple step requires that you have an agenda in the first place! Training your stakeholders to read and understand the agenda leads to more effective and focused discussions. After each meeting, recap the decisions made, action items, and the responsible parties to ensure everyone is on the same page.

Clustering meetings and setting aside large blocks of time for focused work can drastically reduce the time spent on meetings. This is crucial because, as Shreyas highlights, product managers often spend a disproportionate amount of their time in meetings.

Daily and Weekly Routines

To maintain control over your workload, establish a consistent review routine. Start your day by reviewing your meetings and defining the desired outcomes for each. This proactive approach ensures that you enter each meeting with a clear purpose.

Mark important emails and create filters to prioritize your inbox. The same principle applies to Slack channels—star the most crucial ones and mute the rest. Additionally, keep a close eye on your key metrics, dashboards, and alerts every day to stay informed about your product’s performance.

Gathering information about new customers and reading user feedback should also become part of your daily routine. Consider incorporating this into your daily standup meetings to keep the team aligned on customer needs and feedback.

Intentional Calendar Management

Your calendar is more than just a tool for scheduling meetings—it’s a reflection of your priorities. As Shreyas notes, “Your calendar speaks greater truth than your to-do list.” To make the most of your time, create an intentional calendar that balances meetings, deep work blocks, and personal learning time.

Don’t just schedule meetings; schedule your work as well. At the end of each day, plan the next one. Similarly, at the end of each week, plan the upcoming week. Limit unnecessary meetings and add blocks of time dedicated to focused work. This disciplined approach will help you maintain a balance between reactive tasks and proactive, high-impact work.

Leveraging Deep Work

Shreyas also emphasizes the importance of deep work—uninterrupted, focused time dedicated to complex tasks. Schedule “no meeting days” and deep work blocks when your energy is at its peak. Mute Slack and, if possible, change your location to minimize distractions.

The LNO Framework

Another valuable concept introduced is the LNO framework:

Leverage (10x): Activities that have a significant impact on your work.
Neutral (1x): Tasks that maintain the status quo.
Overhead (<1x): Low-impact tasks that can drain your time.

Prioritize activities that offer high leverage, minimize neutral tasks, and avoid overhead whenever possible. We already use this approach within our Product team.

Additional Resources

Shreyas recommends exploring “The War of Art” by Steven Pressfield, which dives into the concept of resistance and how to overcome it. Additionally, the “Shitty First Draft” concept by Julian Shapiro is a powerful tool for overcoming perfectionism and getting your ideas out quickly.

By incorporating these habits into your daily and weekly routines, you can enhance your productivity, make better decisions, and ultimately become a more effective product manager. Remember, it’s the small, consistent changes that lead to significant long-term results.

9 years, 9 lessons

Tue, 06 Aug 2024 00:00:00 GMT

This summer marks my 9-year anniversary at efood—a truly remarkable journey that began in 2015 in Marketing as a Digital Marketing Manager and continues in 2024 in the Product & BI departments as Chief Product Officer.

In this article, I’ll present and explain the most important lessons I’ve learned throughout this journey. These lessons are not linear, prioritized, or specific to Marketing or Product. They are essential professional and personal lessons that have helped me navigate different domains and levels, manage successes, and cope with failures.

#1: Take the hard projects.

As “hard” can be described a project that has one or some of the following attributes:

Has tricky implementation,
Requires multiple departments synchronization,
Has (very) tight deadlines,
The risk of failure is high,
The solution and winning approach is not clearly defined,
Having a successful experience as an interim executive,
Launching a truly innovative feature/campaign/project,
Successfully turning around business performance etc.

Taking on hard projects offers a wealth of experience. First of all, it is one of the best learning experiences. Many hard projects are like “shock moments” that allow you to acquire condensed and valuable experience in a short amount of time. Your problem-solving skills and emotional intelligence will inevitably grow—skills that are very valuable for every modern professional.

Lastly, successfully dealing with hard projects will help your company grow with you as an active contributor and will make your career progression easier. Personally speaking, I don’t care how many years you work in a company; I care about the impact you had in those years. Often, impact is connected with hard projects (though not always). Some people fill Excel sheets for 5 years, while others build products from scratch and fail within a year. I prefer the latter.

#2: Choose your battles

If you were in the office, you would hear me saying this many times a day. Choosing (or better, prioritizing) your battles is a very effective concept for both long-term and short-term solutions. It helps you concentrate your efforts on the most important battles and be okay with losing the ones that are not important to you.

The “choosing your battles” concept improves prioritization, makes you more data-informed, and helps you develop better emotional intelligence. It trains you to identify which projects require data, which require a strategic approach, and which require guts.

I am not a fan of losing. I don’t like to lose (does anyone?). Choosing my battles helped me develop the right perspective; I don’t care how many battles I’ll lose as long as I win the most important ones that determine the overall result.

Imagine you have a pistol with 6 bullets (solutions) and 20 enemies (challenges). Choose wisely how to spend your bullets. Sometimes we “shot” all our bullets at a single challenge because winning that specific one helped us overcome the other 19. You’ll always have fewer bullets than challenges.

#3: It takes two to tango

This one is another cliché in our office and part of our Product principles. It is very helpful because whatever happens, no matter what others did wrong, it is about YOU and how you can become a better professional, even if 99% of the fault is not yours (you can work on your 1%).

There is always room for improvement, and I prefer to focus on this rather than blaming others. It’s very useful to always check what you did wrong in every situation you don’t like because this creates actions rather than excuses.

We humans are very good at spotting the bad moves and correcting others’ actions, but when it comes to ourselves, due to our biases, we struggle to face reality. This principle was mostly inspired by James Clear and his book “Atomic Habits,” which introduces the concept of tiny marginal gains and the idea of becoming 1% better every day.

#4: Do not compromise the edges that matter

Inspired by Seth Godin, this lesson is very helpful both for Product and Marketing. In a world where multiple projects run simultaneously and perfectionism is not an option (I’m not a fan of it anyway), you need to be really careful not to miss or drop anything important in a project, product, or marketing campaign.

Examples:

Adding one more solution to a corner-case flow in a Product Requirements Document (PRD) that has already been reviewed and approved but is missing something that will impact delivery.
Making one more commit to decrease the milliseconds of an API call that is marked as “fast” in your reports but is crucial for the whole infrastructure.
Creating one more design variation to fix the width of a component, even if the fix isn’t visible to the customer’s eye, but is important for design system consistency.

The edges are (almost) infinite. You’ll need experience to successfully recognize the right “edges” for every situation. This experience is developed through feedback loops and retrospective sessions with fellow teammates. Once you gain it, it becomes very useful and makes you very effective.

Any Seth Godin fans here? This is for you.

#5: Manage Expectations

Managing expectations is very important, especially in bigger companies. Multiple projects run simultaneously, deadlines need to be met, and various business performance metrics need to be moved, all within a beautiful “chaos” that needs to be directed in a timely manner.

One good technique is to highlight the things we WILL NOT do. In many projects, we discuss what we need to do and good ideas, and what the competition has, but we do not often discuss the hard NOs for an initiative. For me, specifications that will not happen are equally important because they shed light on areas we have taken full responsibility to avoid working on.

The most effective approach, in my opinion, is one that Holly Sargeant taught me: under-promise and over-deliver.

This doesn’t mean hiding something until it happens. It is a pragmatic delivery approach, where everyone is aligned on what will be delivered, while the team tries to be on time and provide “the cherry on top” where applicable. This is especially relevant in Product Management, where we act as the intermediary between Business and Tech, handling business expectations while working with Tech to deliver the best solution.

#6: Avoid drama & build resilience

Oh god, please avoid drama. Imagine you have energy for 10 thoughts within a day. If you spend your thoughts on drama, the result will be zero. If you spend 5 thoughts on something new and 5 on how to promote and plan it, even if there is no additional capacity to build something new, you train your mind and shape your discovery and planning acumen, rather than your blame game acumen.

Drama is very sneaky. Once you get in, it will drain your energy, affect your mood, and take your clarity away. Drama often arises in hard projects when barriers seem unbreakable, deadlines are missed, complaints arise, or a project fails.

Avoid it at all costs. Focus on impact. Focus on effort. Focus on lessons learned. Focus on feedback. Don’t let drama distract you from getting something valuable. Even in the biggest failure, there is always something valuable.

#7: Lead by example

The transition from individual contributor to manager opened a new era for me. The first (and most difficult) task was to abandon my tasks and delegate them properly. This is the first pill you have to swallow. Many individual contributors fail to become managers due to this.

Having read many managerial books, received mentoring from exceptional managers, and trained and educated in management, my whole managerial framework starts with my core managerial principle:

Lead by example.

Simple but challenging principle.

Do you want a data-driven team? You should be data-driven.

Do you want a world class data-driven team? You need to be world class data-driven professional.

Do you want a disciplined team? You should be the most disciplined professional in the team.

Do you want a team that focus on impact? You should be the first that brings the most impactful projects in the table.

You set the rhythm. You heavily influence the culture. You lead by example.

#8: Continuous learning is a skill

I started in Marketing and eventually evolved into BI & Product Management. But even if I had stayed in Marketing, things change all the time. Tools change, strategies and tactics adjust, new tips and tricks emerge, and AI disrupts every field and sector, especially in digital industries.

The only things that stay the same are the fundamentals. In my opinion, the fundamentals will remain the same even after AI. Owning the fundamentals of Product Management is mandatory if you want to become a good Product Manager.

Besides updating your skills and knowledge, the expansion of knowledge is also important. As you progress in your career, you will face different challenges that require new knowledge (e.g., hiring).

After years of continuous learning, I have come to the following conclusions:

Books are the best source for soft skills (communication, leadership, etc.).
One book is not enough. You need at least 2-3 quality books on a topic.
Re-reading good books every once in a while is underrated.
Courses are very good for acquiring hard skills (coding, design, etc.).
When it comes to tools you will spend a lot of time on, read their documentation religiously.
Make sure to buy updated or fresh courses because tools evolve.
If you don’t practice what you read, you won’t develop the skill.
Practicing a skill requires time. Even if you read the books and practice, you need to continuously work on the skill. I haven’t defined the sweet spot in terms of how much time or how many times you need to do something to become competent, but I think it varies depending on the complexity, importance, and your anticipation of this specific skill.

As a learner, it’s useful to set your learning style. Do you want to become an expert in a field, or do you want to accumulate various skills and build your own unique skillset? Here is a book that will help you define if you’re an expert or a generalist. (I’m a generalist).

#9: Care

Inspired by Seth Godin Godin, a Marketing philosopher, this is the last and most important lesson. It distinguishes good from great professionals. It that can’t be taught or systemized. It that can’t be measured. You won’t find it in any job description. There is no lesson, or course (maybe some books like “Purple Cow”)

It’s really simple, yet rare:

You need to care.

If you don’t care, you will only deliver stakeholder requests.

If you don’t care, you will only deliver your requests while ignoring stakeholders.

If you don’t care, you will deliver something you believe is wrong because someone else believes it is right.

If you don’t care, you will just follow the process.

If you don’t care, you will not respect the process.

If you don’t care, team problems are not your problems.

If you don’t care, team problems are your only problems.

If you don’t care, you will use the hierarchy to hide.

If you don’t care, you will ignore the hierarchy in your own favor.

Care is highly connected with ownership. “It’s not mine to fix” is very common, especially in projects that are fragmented or cross-departmental.

I can’t stress this enough. We spend the majority of our daily time at work. Our time is valuable. Make it count.

Helpful book:

Cheers to the years to come and the challenges and opportunities that will arise!

Made to be hit

Thu, 25 Jul 2024 00:00:00 GMT

In February 2024, I went to a music store to buy drums. This was a childhood dream of mine, and I thought it was the right time to acquire a set of drums. The set is electronic because I live in an apartment building, so I need to make sure I don’t get expelled by other residents due to drum noise!

When I asked the salesman, “Can they withstand multiple hard hits?” he answered, “These things were made to be struck hard and often. This is their job. It’s their job to be struck and create a beautiful result.”

Somehow, I resonated with this as a PM. Product Management emerges as a profession positioned between Business and Tech. Both sectors are known for their pressures but in different areas. Business pushes to generate more revenue, while Tech focuses on building a scalable product with the right documentation and clean code. I am overgeneralizing, but my point is that Product Management was made to handle pressure. This is the job: handling the pressure and creating the best possible experience for the customers.

Since I’m a novice drummer, I asked him again, “How difficult is it to learn drums?” He replied, “Entering the world of drums is very easy. You can learn a specific beat and be ready to start within a month. Becoming good has a steep learning curve, though. This is not the case with all musical instruments. Guitar, violin, etc., have much steeper learning curves before you can play your first song.”

Again, this is the same for Product Managers. Learning the basics is not so hard (compared to Finance, Engineering, Medicine, etc.), but mastering the profession of PM requires a steep learning curve and many years of practice.

I feel that there is a good analogy here.

A book about politics?

Tue, 23 Jul 2024 00:00:00 GMT

I read a book months ago which I found pretty interesting, but something was missing. I didn’t understand what or why, but I felt something was missing. This changed when I discussed it with a friend who had also read the book. We had an extensive discussion, and to my surprise, although the book is called “The 48 Laws of Power,” it’s not actually a book about the laws of power but about politics—specifically job politics or corporate politics.

Politics at work is a heated topic.

Everyone dislikes politics.

Everyone tries to avoid them.

Everyone says they are bad.

Everyone says that a good working environment has zero politics.

Yet, politics exist everywhere!

My point of view is that politics are inevitable. Once your company exceeds three people, politics start to exist. Based on this perspective, you can’t eliminate them. The only thing you can do is minimize them.

Here are a couple of “laws” described in the book that fit perfectly with office politics:

Law 1: Never Outshine the Master.

In each room, there is always a master. Most of the time, it is the manager, the CEO, or a great professional. It’s not wise to outshine the master.

Law 5: So Much Depends on Reputation — Guard It with Your Life.

Reputation matters a lot, especially in the corporate world. For example, this is one of the reasons behind many LinkedIn posts.

Law 11: Learn to Keep People Dependent on You.

By keeping people dependent on you, you seem more valuable than you are, and it may create situations where you are not easily replaceable.

Law: 19: Know Who You’re Dealing With – Do Not Offend the Wrong Person.

For example, offending the decision-maker in a room is not a wise move.

There are 44 other laws in the book. The laws above might not be applicable to your situation. Not all laws will be applicable, which is reasonable because politics are highly customized to your working environment.

Lastly, I wrote this post NOT to encourage you to train yourself in politics and learn to manipulate others. I hate politics because I believe they create noise, waste, and, most of all, a toxic environment. All I’m trying to say is that politics exist, and this book is a good resource to identify them effectively and find ways to cope with or overcome them.

Here is a nice and relevant presentation as well:

Eiffel's Tower | Nickolas Means | #LeadDevLondon

Video sponsor: Formidable (https://formidable.com/) When Gustave Eiffel built his namesake tower, it was nearly twice as tall as the tallest structure on Earth. His crews built it in an astounding 22 months, pioneering new construction techniques to deliver it in time for the opening of the 1889 Exposition Universelle. It was amazing then, and it’s just as captivating today. We all say we want to do groundbreaking work, just like Eiffel, but what does it actually take to push an organization forward? The answer starts long before the work itself. Let’s see what we can learn from how Gustave Eiffel went about building his record-shattering tower. About Nickolas Means Nickolas Means loves nothing more than a story of engineering triumph (except maybe a story of engineering disaster). When he's not stuck in a Wikipedia loop reading about plane crashes, he spends his days as a Senior Engineering Manager at GitHub. He works remotely from Austin, TX, and spends most of his spare time hanging out with his wife and kids, going for a run, or trying to brew the perfect cup of coffee. Find out more: https://london2019.theleaddeveloper.com/talks#nickolas-means Lead Dev is a community for technical leaders. We have a series of conferences and meetups designed with the needs and pain points of technical team leads in mind. We programme each conference around our three key themes of Team, Tech & Tools. The talks are designed to help very practically with common problems in these areas. To find out more about us, visit: https://theleaddeveloper.com/

https://www.youtube.com/watch?v=RNGZTkM2xOU&t=1432s

Universal Analytics: End of an era

Mon, 01 Jul 2024 00:00:00 GMT

As you might already know, Universal Google Analytics services cease to function starting today (July 1st). Google Analytics 4 will replace them.

I spent a significant amount of my professional (and personal) time on Google Analytics (UA). I will definitely miss them, since they were my partner in many (if not all) problems that required data to be solved.

Before I switched to Product Manager, as a professional Performance Marketer, there was no chance to get hired in the efood Marketing team without having professional knowledge of Analytics and understanding their importance for both Customer experience and Marketing performance.

Even in the Product team, for many years, especially the early ones, one of the mandatory skills for EVERY Product Manager that was about to be hired for our Consumer team was the grasp and knowledge of how GA works (technically), why it matters, and how a PM can utilize GA as a tool to create a plan of action.

We ditched that approach now because: i) we are using an internal tool for measurement and optimization, and ii) I missed the chance to hire some very good PMs simply because they were not proficient in Google Analytics.

💡

Featured article in DH blog about conversion rate measurement with UA Google Analytics.

What is changing with Google Analytics 4:

I’m still learning how GA4 works, so I’ll try to describe what I have understood so far. The main difference is the change in the data model measurement regarding sessions and events. GA4 is an event-based tracking service, while UA was a session-based model. One of my assumptions for this change is that it is correlated with the pricing change of AdWords that changed significantly throughout the years, from CPM (cost per mille) to CPC (cost per click) to CPA (cost per action).

On the good side, GA4 supports cross-platform tracking, with better integration between web and app tracking. Moreover, it comes with cookieless tracking capabilities and enhanced user privacy. It also supports BigQuery export for all customers (not only paid), which gives you better flexibility in reporting (but it is more complex).

Guide from Google regarding changes between GA4 and UA

I’m still waiting to understand the difference between data-driven attribution (GA4 default) and last-click attribution (UA default) and if it is better on behavioral flow and multi-channel funnels reports. (relevant old article, regarding attributional models)

One thing I know for sure is that re-learning as a skill is the new black, and GA4 is a very good candidate for all data-driven PMs out there. Sorry, AI tools re-learning is the new black, but still, GA4 is super important.

#1 thing to do when you start a new job

Thu, 13 Jun 2024 00:00:00 GMT

Since we are continuously onboarding new team members, I would like to highlight the top thing (in my opinion) that every newcomer should do in a new job. This approach applies to every new job, not only in product management, as it is useful for every modern professional.

The most important thing that you should do as a newcomer is to understand how the industry of the business works.

If you’re already an industry expert, you won’t need to do much—mostly just adapt and update yourself on the current competition and challenges. This tip is more useful for people coming from other industries.

Most companies and teams have an onboarding process (we do as well) that presents the way they work. A typical process includes strategy & vision, planning processes, collaborative teams, goals & KPIs, and the list goes on. The onboarding process is good, but it will only help the newcomer understand WHAT processes are in place, rather than WHY they are in place. Understanding the industry will help you grasp many important things, such as why you work the way you do, what the main challenges are, the strengths and weaknesses of the competition, and a glimpse of the future based on where the industry is heading.

This approach will help you develop better product & business sense as well.

As a disclaimer, I don’t think anyone will ask you to understand the industry in order to work there. This is an approach for professionals who want to go the extra mile and amplify their impact in their job, especially in terms of product discovery. Here is another tip for success for a newcomer.

To do this effectively, I will break it down into three different sections and add examples from efood.

The Industry Fundamentals

First of all, you need to understand the industry. We are an online food delivery marketplace that operates locally in Greece and Cyprus. Here are some readings about the fundamentals of online marketplaces:

Aggregation theory (by Ben Thompson)
Top 100 Marketplace report (by a16z)
Marketplaces in the age of AI (by a16z)
The concept that explains everything about marketplaces (by Dan Hockenmaier)
20 best essays for Marketplace startups (by Andrew Chen)

Here are some insights from the reading above:

Our prioritization should be balanced across a triangle of:

Segment	Priorities
Customers	Customers want to open our app and find great deals and discounts from all the available restaurants while getting the best user experience without flaws or technical issues in order to place their order. Having low traction with customers will lead to an unwillingness of vendors to enroll in your marketplace and limited opportunities for business revenue.
Vendors	Vendors want to enroll in our service in a smooth and convenient way and get all the necessary guidance and tools from us to help them generate orders and revenue. Having low traction with vendors will limit the options (and potentially discourage) customers from ordering.
Business	Businesses want to generate revenue, keep operating costs low, build a great brand, provide value for both customers and vendors, have social impact, and attract and retain top talent. Having low traction in business revenue will not allow you to attract top talent and will limit your options in terms of promotional actions for both vendors and customers.

The Role Models

Are there any role models in your industry? A role model could be a company that has the same operational model as you but operates on a global level, is highly innovative, or has a great brand, etc. You have to set the role models; there is no default. In our example, DoorDash, Talabat, Baemin, Foodora, and Instacart can be good candidates for role models.

Role models show you the way and the future. You don’t necessarily have to religiously follow them and replicate all their features, but keep an eye on them consistently, especially for A/B tests & new initiatives, for inspiration.

Industry & competition

Since we covered the fundamentals of the industry, we need to understand how the industry that our business operates in works. In our example, efood operates in Greece, so our main direct competitors are BOX & Wolt. If there are indirect competitors, add them as well.

Study them.

Use their services.

Subscribe to their newsletters and social media profiles.

Read their financial statements.

Try to gather as much information as possible. Being on top of your competition will help you understand their strategy, their advantages & disadvantages, how they respond to your initiatives, and why people love (or hate) their brand.

Understanding what your competitor is doing better than you does not necessarily mean that you will copy it. Our main purpose here is to understand the opportunity or the problem, rather than to get the solution.

Questions to answer:

How do they acquire new users?
How do they retain their users?
What are the differences in the operational model?
What are their strong areas?
What are their weak points?
How would you position them against your business?
What are the gaps between them and your business?

Personally, studying the competition helped me a lot in comprehending the dynamics of both the competition and our business. It provided context for things that are very useful but outside of my job scope, and it greatly improved my product discovery skills.

[Hot take] End of Software?

Sat, 01 Jun 2024 00:00:00 GMT

Today I came across an interesting article by Dan Hockenmaier. Dan is a professional in analytics and a writer of many insightful articles on Online Marketplace theory. You can find a couple of his articles in my knowledge base.

💡

In my knowledgebase you can find a lot of interesting articles from me and my network.

Today, he wrote an essay about the death of software. Here is the link. Since it’s a Google Doc, I will copy the original content here as well, but please check the original link in case something has changed.

Original content

To understand how software will change, we can benefit from studying how technology has changed other industries. History tends to rhyme, if you listen.

Before the internet, media behaved very differently—it was expensive to create. You had to pay people to make content, edit it, and distribute it. Because content was expensive to create, it had to make money. And consumers paid—newspapers, magazines, books, cable, and pay per view. Warren Buffett famously loved newspapers—and who wouldn’t love a predictable subscription business with local monopolistic dynamics?

When the internet happened, media companies viewed it as a way to reach broader audiences and reduce their distribution costs. But what no one saw coming was that the internet not only reduced distribution costs to zero, but it also drove the cost of creating content to zero. User generated content flourished, and when content doesn’t cost anything to create, it no longer has to make money. How does content behave when it no longer has to make money? The relaxation of this economic constraint led to a Cambrian explosion–you can take a picture of a cup of coffee, post it to a million views or none at all and the market clearing price is still met. This produced a deluge of content that none of us could reasonably consume. This necessitated products to direct attention, merchandise this content, and route us effectively–we understand these now as user-generated content platforms.

These platforms completely T-boned media companies. As a media company, you were competing for the same attention of users, but with a strictly higher COGS. The more people you had on your payroll that were creating content, the more exposed you were to being flanked by user-generated content platforms. Structurally, investing in media has been a losing value proposition ever since and value creation has shifted entirely to the platforms that control distribution.

Software is expensive to create. You have to pay people to create it, maintain it, and distribute it. Because software is expensive to create, it has to make money. And we pay for it–software licenses, SaaS, per seat pricing, etc. Software margins have historically been an architectural envy–90+% margins and zero marginal cost of distribution.

Software is expensive because developers are expensive. They are skilled translators–they translate human language into computer language and vice-versa. LLMs have proven themselves to be remarkably efficient at this and will drive the cost of creating software to zero. What happens when software no longer has to make money? We will experience a Cambrian explosion of software, the same way we did with content.

Vogue wasn’t replaced by another fashion media company, it was replaced by 10,000 influencers. Salesforce will not be replaced by another monolithic CRM. It will be replaced by a constellation of things that dynamically serve the same intent and pain points. Software companies will be replaced the same way media companies were, giving rise to a new set of platforms that control distribution.

SaaS, ARR, magic numbers–these are all shorthand to understand the old model of business building in software, one where the expense associated with creating software was a moat. The invisible hand has been stayed in software for a long time, but LLMs will usher in its swift, familiar corrective force. Majoring in computer science today will be like majoring in journalism in the late 90’s.

My 🔥 take on this (briefly):

I could write an entire essay on this, adding more examples and explanations, but here is my quick take. I’m not sure about the end of software, but I am pretty sure about changes in the software industry.

Let me explain:

NVIDIA’s CEO mentioned in one of his presentations that AI will soon enable everyone to develop indie programs by translating natural language into code. There are already some initiatives and startups in this field, but it’s still immature.

If that happens, literally everyone will be able to create lightweight programs to solve small or relatively small problems, while API integrations will allow those programs to be compatible with each other. Even today, with a little coding knowledge and the support of co-pilot, you can create a simple Chrome extension on your own. Try it.

The above statement does not mean that existing software companies will die. They might face bigger competition and need to change their pricing approach, etc. They won’t necessarily die; they’ll need to adapt to the new era. Also, a software company is not only about software and code. It is about customer support, brand, network effects, and data privacy that create a holistic experience and perspective for a software company.

The analogy between media and software is relevant but not 100% accurate, in my opinion, because software is a broader industry than media. The difference between consumer and enterprise software is super accurate, though. I created my own application, but I can’t utilize it in the company I currently work (efood) due to corporate restrictions.

Surely, AI is a fundamental evolution that will shake and disrupt many industries, but we are still in the midst of this process and cannot predict exactly how AI will evolve in the next year (ChatGPT 3.5 is less than two years old 🤯).

From Marketing to Product and vice versa

Tue, 23 Jan 2024 00:00:00 GMT

Recently, there’s been quite a buzz about Brian Chesky’s latest interview in Lenny’s newsletter. They covered a myriad of topics, but one, in particular, resonated with me the most. Chesky’s approach to integrating product management with marketing at Airbnb focuses on blending the creative and strategic aspects of both domains. He believes that understanding a product and marketing it are intrinsically linked—a product manager must be adept not only at creating the product but also in communicating its value to the market.

This integrated approach ensures that the product is not only well-designed but also effectively positioned and marketed, aligning with Airbnb’s mission of creating unique and reliable customer experiences. Chesky’s strategy represents a shift from traditional siloed roles, emphasizing a more holistic and collaborative approach to product development and marketing.

This resonated with me because, before switching to Product Management, I was a Marketeer. I can intimately feel the connection between these two worlds.

Although there is debate about this approach, I agree with it. That’s why I decided to write this post, highlighting what a competent Performance Marketeer can bring to the Product Management table.

Before going deeper, let’s consider this enlightening graph from Seth Godin’s blog about the circles of Marketing. Can you guess which circle lies at the very core?

Read more here

USER INTENT

The main and most important skill valuable for both PMs and Marketeers is deeply understanding the intent of the user. User intent refers to the goal or purpose a user has in mind when typing a query into a search engine or when interacting with a product or service.

💡

Marketing strategy 101: The right message, on the right users, on the right time.

Understanding user intent is crucial for the following reasons:

User intent defines the type of communication

Different intents require varied communication channels and concepts. As a PM, you might work with product personas based on product usage (e.g., a power user of feature X) and customer segmentation (e.g., a churned user). However, as a Marketeer, utilizing marketing channels to match customer intents is key.

For instance, your power user might be active on TikTok or LinkedIn—platforms that demand completely different communication approaches. Developing holistic personas enables efficient and resonant communication, which is invaluable for any company.

User intent defines the search query

Consider the difference between searching for “souvlakia halandri” and “souvlakia halandri efood.” The intent behind these queries differs significantly. The first query suggests a general interest in souvlaki in Halandri, while the second implies a preference for ordering souvlaki via efood.

This distinction impacts several aspects:

User Perspective: The intent varies between general exploration and a specific service. On the first search user might be unaware of our service, while on the second one, most probably the user is an already active customer.
Ads Cost and Performance: The cost and performance of ads differ significantly between these queries. On the non-brand query the cost might be 10x more than the brand-query, while the performance of non-brand is 10x worse than the brand-query performance.
Organic Search Results: The nature of organic results changes as well, since the non-brand query includes various results while the brand-query includes only brand-relevant results.

“Souvlakia halandri” results:

“Souvlakia halandri efood” results:

User intent defines the landing page

The reason behind having multiple landing pages in a website is that each landing page, was developed to match a user intent. Landing pages that match effectively the intent of the users, rank higher in Google search and generate incremental visits to the Product.

This is not a simple skill to develop.

It’s called SEO and there are multiple tools and professionals that are working solely on this. The whole industry of SEO was build around this concept, an industry that has 45 billion USD market size. Since we talked about SEO, there is a whole section in SEO called on-page SEO that is considered as “technical” since it includes technical configurations in the product page.

Here is a nice article by Skroutz that talks about SEO and user intent. Skroutz has one of the most interesting SEO strategies in Greece.

Also, regarding intent and landing pages, here is an interesting article from Jeff Chang that talks about the intent/ease framework for conversion optimization. In this article, Jeff talks about the high and low intent of a user to convert and correlates it with the type of landing page.

User intent defines the UX

The overall experience of a product should be driven by user intent. Every user that lands on your product wants to do something. If you manage to understand and provide it, then the probability to convert increases dramatically. This is a double-edged sword, though, because if you don’t serve the intent correctly, the conversion will not happen.

Here starts the tricky part.

What if the intents that your product serves are multiple?

Take JIRA, for example. JIRA is being used by multiple professionals across industries and roles. PMs, agile coaches, delivery leads, engineering managers, engineers, QAs, Product & Tech directors are some of the professions that are using it, all of them having different intents.

Examples:

PMs use it to write specifications,
Agile coaches to measure the velocity & effectiveness of the teams,
QAs to find the acceptance criteria of the feature that is about to be launched,
Delivery leads to monitor roadmap milestones, etc.

How do you handle when multiple intents should be served within one page?

A good example is efood. There are customers ordering souvlaki, pizza, groceries, flowers, drinks, etc., while their intent is one of the following:

Order because they are hungry,
Order because they need caffeine,
Social-eat due to an event (birthday, football match),
Order with a specific amount of money in mind,
Order something that will arrive quickly,
Order what others are ordering because they are not sure,
Order from their favorite restaurants,
Order groceries for tomorrow,

In order to handle multiple intents, a common practice is to group them around specific concepts. For example, in a marketplace, the two biggest user intents are: i) convenience, and ii) discovery.

Convenience is about ordering your favorite food from your favorite restaurant in the easiest and most convenient way. The purpose here is to allow customers to order as fast as possible using shortcuts, such as the reorder swimlane or the vertical swimlane, or even search (that can be used either for convenience or discovery).

When it comes to discovery, customers want to explore additional options besides the regular ones (that can be found in the reorder swimlane), and here comes the discovery page, where customers can see vendors around different concepts, such as:

Affordability and how to find discounts (Pinata),
Personalized suggestions based on your behavior (picked for you),
Groceries and items that you can find in efoodMarket (Weekly deals),
Social proof and restaurants that are popular in general (Lovable brands),
Curated vendors (Red Selection),
New vendors to try (New restaurants),
Social proof and popular local shops (Popular vendors),
Top rated specific cuisine (Top rated burgers),
Last but not least, you see some XL banners that communicate multiple initiatives as well, from a CSR initiative to an offer bundle of a specific vendor that you should not miss.

efood app home above the fold

efood app home swimlanes

A good PM should always strive to satisfy the user’s intent while also meeting business objectives and considering partner satisfaction, combining all these aspects into a balanced and smooth experience. Try different products, and you’ll be able to understand which ones handle user intents well and which ones are struggling.

CONVERSION RATE OPTIMIZATION (CRO)

Another skill that a competent performance marketeer is skillful at is CRO, or conversion rate optimization. Conversion Rate Optimization (CRO) is a process of enhancing your website and content to boost conversions, which can include making a purchase, signing up for a newsletter, or any other action you want your visitors to take. It involves understanding what drives, stops, and persuades your users. CRO actually lies between Product and Marketing, and it was one of my core skills as well.

To be good at CRO, you need proficiency in the following skills and concepts:

Product performance has high correlation with Marketing performance

Yes. Product performance correlates heavily with Marketing performance. In efood, paid traffic, which means sessions and users that are coming from marketing campaigns, has a significant % (double-digit) of our traffic overall.

Now imagine that the marketing team has the wrong targeting on their campaigns. The result will be the generation of low-quality sessions that will lower the conversion rate of your product.

On the other hand, even the slightest increase in the conversion rate of your product (due to a product feature or user flow optimization) will definitely save marketing budget and increase the performance of the respective marketing campaigns.

Targeting	Product flow	Performance
Targeting wrong people	Bad user flow	Bad
Targeting the right people	Bad user flow	Mediocre
Targeting the wrong people	Optimized user flow	Mediocre
Targeting the right people	Optimized user flow	Best

Google analytics

GA and other relevant tools that measure user behavior are a must for CRO. There is no CRO without qualitative and quantitative data. Many people consider Google Analytics a marketing tool. Although I disagree with that, I understand its significance within a performance marketing team. Also, I don’t know any competent performance marketer who is not proficient in Google Analytics, while there are senior PMs out there who are not even aware of how analytics tools work.

Segment or die

A very popular concept in performance marketing is segmentation. There are many user segmentation concepts, such as RFM, but being able to identify the right user segments and connect them to product solutions and challenges can be an unfair advantage, especially in consumer products with millions of active users.

Performance marketers are able to handle datasets on their own, making all the necessary segmentations because there is no other way to understand if a campaign is underperforming or if a specific campaign is lacking in a particular user segment (e.g., mobile users) that needs tweaking in the campaign settings to work.

ROI & incrementality

Similar to incrementality, measuring ROI is a core skill for a performance marketer. During the years, we spend millions of euros on performance marketing, so as a team, we needed to know the pragmatic return on investment. There were other metrics more ad-friendly, like ROAS (return on advertising spend), but ROI provides a holistic view, an approach that is essential for every multi-channel marketing strategy.

The holy grail of marketing is measuring incremental performance. Incrementality includes all the orders that came additionally from the orders that would have come organically either way. Incrementality is important for product management as well, since it helps identify not only the adoption of a feature but also the incremental value it adds to the product.

Here is an interesting articleabout incrementality and marketing channels by Avinash Kaushik.

Prioritization

I truly believe that the next best profession for prioritization, besides Product Management, is Marketing. The reason is simple: in most modern businesses, the most important and/or scarce resources are the marketing budget and the technical resources.

A Performance Marketeer transitioning to Product Management will be very familiar with the prioritization process, including assessing the opportunity cost, which is crucial for an effective prioritization process.

After all, everyone has a Marketing idea and a Product idea. I’ve never seen someone had a Finance idea, or a legal idea 😎.

A/B testing

A/B testing is part of the core skill set of a Performance Marketeer. Even though many campaigns have become automated and a black box by the big advertising players, A/B testing is embedded as a mindset.

There is a Marketing book that was written back in 1923 (!) called “Scientific Advertising” by Claude C. Hopkins (a must-read for every Marketeer) that talks about the importance of experimentation.

Copies, creatives, budget, targeting, goals, everything is a good candidate for experimentation in Marketing. Here is an interesting test that we did 6 years ago, which might be resolved currently, but the mindset behind it should remain.

Last but not least, I wrote a book about Marketing on 2020 that includes tips for Agile Marketing and how to operate and scale a Marketing team plus a detailed guide on a growth framework we used back in time, inspired by Growth Engine by Brian Balfour. Here is a brief overview:

Read:

This is the first and most important step in the process. “Read” means that you have to check your data. In order to become user-centric, we worked hard and managed to build a foundation of user metrics and user feedback so that we could avoid our own personal and professional biases and focus solely on our user’s needs. This was a decision that we made firstly as a business and secondly as a department and took months before it started working efficiently. We have a couple of sophisticated systems in place enabling us to have a comprehensive view of the impact of our efforts. Please note, considering yourself as “data-driven” does not necessarily mean that you evaluate everything with data

Learn:

This is the second step and it can be translated as “understanding the problem/opportunity.” This step needs creativity, observation, and anticipation. Once we have “read” the insights, we need to understand what happened and plan our next steps accordingly. We know how users came to our product (thanks to analytics), where they landed, and what pages they navigated along with their exit pages, but we are missing something. We are missing the why.

Test:

The third step is to test the solutions that you have come up with. This can be an A/B test, a split campaign test, a user testing session, etc. Testing is a science by itself. We feel that sometimes testing and experimentation are a little overrated at least in our industry (after more than 100 tests we are not even close to analyze how an A/B test in an ad copy can produce tons of additional orders), but we find value in it and use a couple of testing approaches to cover all our needs.

Validate:

This step is dedicated to the validation of proposed solutions. For validation, we mostly use data performance, wherever applied, or our experience in the business. We use a wide range of validation techniques, from statistical significance indicators to plain positive performance results or user testing insights.

Repeat:

The final step is to replicate every success you had and digest the lessons you learned from the previous steps. In the book, I briefly describe how we utilized this framework for our YouTube channel, plus there are plenty tips and examples for each step.

I would like to close this article by sharing an old guide of ConversionXL about CRO. For all those that are interested in CRO, feel free to check it.

Having all the above into consideration, I feel that Product Management could emerge as a very interesting path for a competent performance Marketeer. Of course, there are pitfalls as well (e.g., lack of technical competency), but having the above skills in place can act as a great kickstart to start as a PM and work to develop the missing skills too.

To my friends that work as Performance Marketeers:

Join, the dark side. It’s fun.

A relevant video from a meetup we did with George discussing Product & Marketing topics.

[Guide] Product hiring framework

Thu, 30 Nov 2023 00:00:00 GMT

Here is a framework that I used to hire Product people across different disciplines (design, analytics, BI) and levels (from juniors to seniors). It is based on an article from Marc Andreessen’s blog that was posted 6 June 2007.

Sidenote: Read old blog posts. I find a lot of value in old articles. I find them less sales-oriented, with deeper meanings and value that lasts. This is not applied to all but there are some great old articles that you should not miss. Check my knowledge base and you will find plenty of them.

Inspired by this article:

Pmarchive · How to hire the best people you've ever worked with

There are many aspects to hiring great people, and various people smarter than me have written extensively on the topic. So I'm not going to try to be comprehensive. But I am going to relay some lessons learned through hard experience on how to hire the best people you've ever worked with-particularly for a startup.

https://pmarchive.com/how_to_hire_the_best_people.html

How should I choose the right candidate?

In order to choose the right candidate, you need to define what you really want. You should take into consideration the following:

[Desired professional level] Do you seek junior, middle, senior or leadership position?
[Mindset & Sense] Do you seek someone to focus on commercial & customer problems? Someone to optimize your planning process or someone to get shit done?
[Skillset] Does the team need someone to inspire them, help them deliver more or give them the right direction?
[Business maturity] Is this a new business vertical? A mature business function? A bet?
[Team maturity] Is this a team full of experienced developers? Does the team have designers?
[Stakeholders maturity] What does the business expect from Product team? To deliver more? To deliver better solutions?

Once you have all the above clarified, you can develop a profile that consists of personality traits & desired skills. Based on this you will craft your job description.

You have to choose no more than 9-10 key areas. The reason behind this number, is that the more the areas, the “thinner” the examination.

Since interview time & resources are limited, you have to focus your energy on specific areas that you would like to explore, rather than randomly checking facts & past experience from the candidate.

[Personality] Hiring Pillars

These questions should not necessarily be answered directly. It’s better to ask for specific examples around these topics. Examples can be based on past experience or can be theoretical based on recent trends or best practice. The examples will help you understand if the candidate has the right mindset & attitude in place.

Drive

➕ Do you want to become top PM%?

➕ Are you disciplined enough?

➕ Are you motivated enough?

Curiosity

➕ Do you have the necessary analytical thinking skills to take the right decisions?

➕ Do you love your craft?

➕ Are you creative enough?

➕ Did you just launch a product feature or you are proud of it?

Ethics

➕ Do you like transparent communication?

➕ Do you have the resilience to overcome potential obstacles without whining and drama?

➕ Are you a team player?

➕ Are you a culture fit for the team?

💡

If you are super-scrupulous about your hiring process, you’ll still have maybe a 70% success rate of a new person really working out—if you’re lucky.

And that’s for individual contributors.

If you’re hiring executives, you’ll probably only have a 50% success rate.

That’s life. (source)

[Skillset] Hiring Pillars

Product Sense

➕ Do you (deeply) care about your customers?

➕ Do you care about UX? What’s the best product out there and why?

➕ Can you collaborate with designers?

➕ Are you data-informed?

Analytical Sense

➕ Can you plan effectively?

➕ Can you prioritize effectively?

➕ Can you manage your stakeholders?

➕ Can you prioritize your time?

Execution Sense

➕ Do you have the right technical skills?

➕ Are you a builder?

➕ Do you care about delivery?

➕ Can you collaborate with developers?

[Leadership] Hiring Pillar

➕ Do you like managerial or individual contributor activities?

➕ Can you effectively manage conflicts?

➕ Do you care about your people?

➕ Can you craft & implement the Product strategy of the next day of your product?

In summary, this comprehensive framework for hiring product professionals is a strategic blend of industry insights and practical considerations, influenced by Marc Andreessen’s wisdom and personal experience.

It emphasizes a meticulous approach to defining candidate requirements, focusing on a blend of personal traits, skillsets, and alignment with the company’s maturity levels in various aspects. By limiting focus areas to a manageable number, the framework ensures a thorough evaluation process, tailored to find individuals who not only meet the job’s technical demands but also resonate with the organizational culture and ethos.

Agile Marketing chatbot

Mon, 20 Nov 2023 00:00:00 GMT

On November 6, 2023, OpenAI introduced custom versions of ChatGPT. These versions blend instructions, extra knowledge, and various skills.

Concept

This is about creating a chatbot to assist with Agile Marketing challenges, drawing from my book on Agile Marketing.

Agile Marketing: From Waterfall to Water Flow

Agile Marketing: From Waterfall to Water Flow : Giamalis, Konstantinos: Amazon.de: Books

https://www.amazon.de/-/en/Konstantinos-Giamalis/dp/6180023735/?&_encoding=UTF8&tag=kgiamalis-21&linkCode=ur2&linkId=54f5ce18a7a0968304ee36533d833875&camp=1638&creative=6742

Setup

Setting up is straightforward. You can either chat with ChatGPT for guidance on setting up a custom ChatGPT or configure the chatbot directly.

Main sections include:

Name:

Agile Marketing Chatbot

Description:

Responds to questions based on the book: “Agile Marketing: From Waterfall to Water Flow.”

Instructions:

The Agile Marketing Chatbot aims for clear and accurate responses. If a question is unclear or lacks detail, it asks for clarification rather than guessing. It avoids making assumptions, ensuring responses suit the user’s real needs. The chatbot focuses on delivering precise answers based on the book, especially on the Nero pillars and Growth engine.

Conversation Starters:

Why should Marketing become Agile?
What is Nero and how is it different from Scrum?
How is the Growth engine used?
What are the benefits of Nero?

Knowledge:

A Google Doc containing the entire text of my book.

Capabilities:

❌ Web browsing

❌ DALLE image generation

❌ Code interpretation

Result

The chatbot is ready to use. The only downside is it requires a paid ChatGPT subscription. I thoroughly tested it, and it’s quite impressive.

Enjoy:

chat.openai.com

https://chat.openai.com/g/g-DSkWHzxr6-agile-marketing-chatbot

P.S.: It’s amazing how simple it is to set up a custom ChatGPT chatbot. Soon, anyone will be able to create their own chatbot, no matter their technical know-how. So, let your creativity flow and try out your wildest ideas!

Ways to Become a Product Manager #1

Sun, 19 Nov 2023 00:00:00 GMT

In this series of blog posts, we’ll explore the best ways to become a Product Manager, drawing on my career and conversations with other PMs.

The challenge in becoming a PM lies in the absence of a traditional or clear path. Most professions follow a standard route, typically involving:

A bachelor’s degree for foundational knowledge (e.g., Marketing),
A master’s degree for specialization (e.g., Performance Marketing),
Starting as a junior in the chosen field,
Career progression.

However, in Product Management, there’s no set path. Anyone can become a PM, and there are many routes into the field.

Here’s my favorite approach:

Build a Product

Building is a mindset. You need to integrate this into your thinking. The best way to do this is to build products.

Develop projects, extensions, simple SaaS, online marketplaces, e-commerce websites, or even games. If you’re not technically inclined, use no-code tools or CMS platforms that don’t require coding.

I’ve built simple products using my coding skills (with a bit of help from ChatGPT) and also used various no-code tools. Both experiences were enjoyable!

If you’re unsure where to start, here’s a roadmap:

Choose a Problem You’re Passionate About

Identify a problem that you care about: Look around for everyday problems or challenges you encounter. Solving a problem you already have increases the chance of success.
Scope and Feasibility: Ensure the problem is manageable. It should be complex enough to challenge you but not so vast that it becomes overwhelming.

Plan Your Tasks

Project Management Tools: Familiarize yourself with tools like Trello or Asana. These tools help in organizing tasks, setting deadlines, and tracking progress.
Task Breakdown: Divide the project into smaller, manageable tasks. This includes market research, defining target users, designing, developing, and testing the product. If you don’t know all the tasks, start simple with your first task, then add the next one, and so on.
Setting Milestones: Establish clear milestones to measure progress. This keeps the project on track and provides a sense of accomplishment. Basic milestones could include defining the scope, developing the product, and launching it.

Note that you need to avoid overplanning. Planning is a supplementary action in your product development, not the primary one. The main focus should be on building your product.

Map the User/Customer Journey

User Personas: Create detailed user personas to understand your target audience’s needs, preferences, and behaviors. If you don’t know any personas, describe yourself, since you are a potential customer for this product.
User Flow: Design the sequence of actions a user will take in your product. This includes navigating through the interface, using features, and reaching their goal. If you don’t know where to start, Pageflows is a useful resource.
User Features: Describe the essential user features mapped in the user flow above. Typical features include login/signup, payment, input forms, etc. There are standalone solutions available that fulfill these needs without the need to build something from scratch.

Choose (or Create) the Design & UX

Learn Design Principles: If you’re new to design, start with the basics of visual design, user interface (UI), and user experience (UX) principles.
Design Tools: Explore tools like Figma. These tools offer functionalities to create and prototype designs. Plus, you’ll experience firsthand the challenges of a product designer.
Buy a Design Template: If you are not interested in investing more time and effort in design, you can buy a design template and proceed with the product design, both fast and affordably.
Developing Your Copywriting Skills: UX isn’t only about design. You need to add compelling copy as well. Use your common sense, your imagination, and ChatGPT to come up with copy that helps users understand and delights them.

Build the Product

Technology Selection: Choose the right technology stack based on your product’s requirements. This could be a coding language (e.g., Python) or a no-code platform. No-code is easier, but learning the basics of coding is more valuable.
Development Process: If you’re coding, learn the basics of the necessary languages. For no-code, explore platforms like Bubble or Webflow.
Testing and Iteration: Regularly test the product for bugs and usability issues. Debugging is a critical part of development.
Tracking & Measurement: Set up a tracking tool. A good combination is Google Analytics and Google Tag Manager. You need to track user behavior and product performance.

Gather Feedback

Let your friends test your product and provide feedback. Also, since you are a potential customer, make multiple tests on your own as well.

Document Everything

Record your development journey. This documentation is now part of your portfolio. You can present it as a blog post, a PDF, or in any other format you like. Include all steps, resources used, and lessons learned.

Announce It Publicly

Launch your product! Three things can happen:

People use and like your product.
People use it but don’t like it.
People don’t use it at all.

Regardless of the outcome, you’ve built your first product.

Here is a relevant example of mine:

kgiamalis.co

/blog/step-by-step-how-to-build-a-chatbot-using-openai-langchain-streamlit

While not a PM’s daily job, this process exposes you to aspects of product work, helping you discover your preferences. Plus, you can continuously modify your product, and there’s no limit to the number of products you can create.

A Beginner's guide on how to use Tag Manager in GA4

Tue, 03 Oct 2023 00:00:00 GMT

This is a short guide for everyone who wants to set up an event in Google Analytics 4 using Tag Manager. This isn’t an advanced guide on how to track complex events on your website, but it’s very helpful for all PMs who want to track basic events without utilizing technical resources.

No data layer, no server-side tagging, just plain event tracking. Our goal here is to help you dive into the tag manager realm with the simplest tracking concept available in Tag Manager.

Before we proceed let’s explain what Tag manager is and why we need it:

Tag Manager serves as a bridge, enabling you to seamlessly integrate and manage tracking codes (tags) on your website without continuously modifying the website’s code. It’s crucial because it offers flexibility in tracking and analyzing vital user interactions without intensive coding or development.

I’ll guide you step by step, providing explanations and details for each step. The objective of this guide is to create a goal in GA every time a user clicks on the “Subscribe” button.

Open Google Tag Manager

Go to https://tagmanager.google.com/ and create your account.

Create your account

Navigate to your account and select “Create your account”. If you have a pending invitation, it will appear at the top, just like the one shown in the screenshot below.

In the setup process, you’ll need to input a name, choose a country, and specify the setup type. I selected “web” since I’m tracking an event on my website.

Make sure you agree to the terms and conditions before creating your account. Once your account is set up, you’ll see the following screen:

Install tag manager in your website

Click on “Admin”, and a settings screen will appear.

Next, select “Install Google Tag Manager” and follow the installation guidelines. Here’s a useful link. Installing Tag Manager can vary depending on the CMS of your website. Generally, it’s straightforward, but you might require some technical assistance for this step.

Add Google tag

You must set up the Google tag before proceeding. Go back to “Workspace” and locate “Tags” in the left-hand menu. Here’s the tag’s entry page:

Move to “Tag Configuration”, and you’ll encounter various tag types. Click on Google Analytics, and a screen will pop up guiding you to utilize the Google tag and GA4 event to activate measurement for GA4. Opt for the Google Tag.

Within the Google tag, input the tag details of your Google Analytics. There are detailed instructions on where to locate it. However, if you face any difficulties, use this path in Google Analytics:

Open GA → Go to Admin → Navigate to Property → Open Data Streams → Access Stream Details → Copy Measurement ID.

Retain the default trigger settings, which is - Initialization on All pages.

Save your changes and return to your workspace.

Make your first submit

Click on “Submit” in the top right corner and publish your initial changes. Describe the changes you’ve implemented on the subsequent page for efficient versioning.

Click “Publish” and “sing like no one is listening”.

Inspect the website

Prior to setting our triggers, we must inspect our website to discern what we can track and how to go about it. Return to the workspace and select “Preview” on the top right. Enter the URL of your website when prompted, then click connect.

A new window will showcase the website you provided along with a message regarding the tag manager connection. If your tag manager isn’t connected, revisit and double-check your installation.

You will also see the following screen which includes all the data you can utilize for your tracking. As you can see my Google tag fired properly and I can see that in tags section.

Every action you undertake on your website’s window will generate additional activities on this page. For instance, if I navigate to a page, a “click” action will be recorded, encapsulating all associated data from that click. For our intent, we’ll focus on the variables section.

Here we can see all the variables captured. It seems that I clicked a button that had text “biography”. There are a lot of variables that you can utilize but for our example you can stick to the basic ones that are built-in and don’t require additional technical development.

As a disclaimer here, tracking an event by using the text on it is bad practice. It’s not recommended because, there might be other buttons that include the same text and the text might change in the future, so you are going to lose the tracking. Tracking clicks based on click-id is better, but for our example, we’ll stick to text. But for a beginner, this is the first step to understand how Tag manager works.

Next, proceed to select the button you wish to track, in this instance, the “subscribe” button. After clicking, you’ll observe the following variables:

I can see here that the click text “subscribe” exists in my button so I’m going to use it in Tag manager.

Setup the trigger

Navigate to “triggers” and click “New”.

Provide a name for your trigger, like “click_subscribe” for the subscribe button clicks. Within trigger configuration, you’ll encounter various trigger types.

More often than not, I opt for click → All elements or just links. You’ll then need to specify which click to track. Select “some clicks” and use the Click Text you identified during the inspection.

By saving, you’ve now established your first trigger. Although using click text isn’t the best approach, for the sake of simplicity in this tutorial, we’ll continue with it.

Setup the tag

Now, let’s configure the tag. Return to the tags section and add a new one. During the configuration, choose “Google Analytics” followed by “Google Analytics: GA4 event”.

Use a consistent naming pattern for your tag, such as “GA_click_subscribe”.

Enter the same Measurement ID you used earlier and define your event name. Again, employ a naming convention that aligns with your requirements. I’ve opted for “click_subscribe” as the event name.

A note here, if you don’t want to measure every hit on a button, you should go to advanced setting and check the available firing options. In my case, I want to measure it once per page, so I will set the right option from the list.

Next, we need to add the triggering. Click on the tab and you will see all your available triggers:

Use the “click_subscribe”, the trigger you created for this purpose. Now you have a proper tag in place that contains all the necessary information.

QA

Before advancing to GA to configure our goal, we need to validate if the tag operates correctly. Revisit the workspace and click “Preview”. We should notice our newly-added tag, though it hasn’t been “fired” yet.

To “fire” it, execute the action that corresponds to its trigger. For this guide, that means clicking the “subscribe” button on our website. If it successfully fires, you’re set to move on.

It fired so we are good to go.

Create GA4 goal

Now, let’s fashion our goal in GA4. Open Google Analytics → Admin → Property → Conversions and click “new conversion event”. Enter the event name precisely as you did in Tag Manager and designate it as a conversion.

Live QA

As GA4 requires some time to disseminate our data, I frequently examine event tracking via the live preview of GA4. Navigate to Analytics → Home → View Realtime → Event Count by Event Name (screen) or Conversions by Event Name.

In conclusion, we’ve successfully crafted our initial tracking event in GA4 independently. This process can be replicated for other functionalities like a “click to call” CTA, or delve into the Tag Manager documentation to establish tracking for additional elements such as IDs, scroll depths, URLs, or even conversion funnels.

Happy tracking!

The first PM

Mon, 25 Sep 2023 00:00:00 GMT

Today, I had a conversation with another product manager about the experience of being the first PM. While I wasn’t the inaugural PM at efood (that was Jason), I did get involved with the product at an early stage. At Instacar, I was the first hands-on PM consultant.

Every popular app or online service we use today, regardless of its size, started with a small team of engineers and the first PM. Many companies have CEOs who built the initial product from scratch, as was the case with efood. Nevertheless, the first PM was always close to this stage.

Moving from zero to one is one of the most challenging stages in product development. The development uncertainty, the volume and complexity of decisions, and the absence of structure can make this journey quite a roller-coaster ride.

In an environment where there is no established rulebook, the first PM must operate without the advantage of set documentation and procedures. Their approach is more practical and agile, always ready to adapt and pivot as required.

💡

The first PMs of successful products are invaluable sources of insights, wisdom, and lessons.

Their experiences, including both failures and successes, form a solid foundation of knowledge that can guide future PMs and other professionals in the field.

Therefore, it is beneficial to seek them out, engage with them, and ask them questions. You might gain insights that could help your product become the next big thing.

It’s all about managing the chaos, thriving in uncertainty, and having the foresight to build from nothing to something.

So, what if you are the first PM within a company?

As the first PM, your role is to set up the foundations of the product development process for the company. To do this, you need to work closely with both Tech and Business. Ensure you cover all important elements of each pillar in Product Management (discovery, planning, delivery). Let’s overview:

On Product Discovery:

Ensure you have an easily-followed, lightweight process to:

Identify the most important customer problems via feedback or user research process.
Identify the most interesting solutions and opportunities in the industry, considering role models, business competition and market analysis.

The main purpose is to match customer problems with feasible solutions that make sense for your business model.

On Product Planning:

Align all functions and stakeholders on what you’re going to build next.

Use strategic thinking to develop your Product strategy and set the right objectives. Huge plus here is to add the things or features that you are NOT going to develop. Also, make sure that you’ll include experiments.
Use quarterly OKRs to match product development with business metrics.
Use analytical thinking and data to properly assess opportunity cost and establish a culture where everyone’s ideas are welcome but also open to debate.
Assess your team resources and draft a pragmatic roadmap that should be regularly checked and adjusted.

On Product Delivery:

Ensure everything is in place to maximize the team’s output.

Create a team structure to minimize internal dependencies and blockers.
Set up a relevant and lightweight development process, retaining only necessary steps for effective and continuous development.
Have all the technical skills needed to be an active contributor to the team.

The way you shape the foundations will define the team’s future culture. Be pragmatic, understand the dynamics of the team and business, and work towards establishing a culture of a product-led organization!

Two Books Worth Revisiting Every Year

Fri, 22 Sep 2023 00:00:00 GMT

One tradition I’ve started in recent years is to revisit books that have made a significant impact on me. Every September, after returning from vacation, I engage in this ritual to mark the beginning of my new season, which for me starts on September rather than December.

Although I have a long list of books that I find valuable, there are two in particular that I read again each year without fail. These books stand out because they continually offer fresh perspectives and insights, even with repeated readings.

INSPIRED by Marty Cagan

Summary: This book provides an in-depth look at the product development process, revealing how successful tech companies create products that customers love. It’s filled with actionable advice on everything from understanding customer needs to effectively collaborating with cross-functional teams. I believe that this is the most valuable book for Product management out there.

INSPIRED: How to Create Tech Products Customers Love (Silicon Valley Product Group)

INSPIRED: How to Create Tech Products Customers Love (Silicon Valley Product Group) : Cagan, Marty: Amazon.de: Books

https://www.amazon.de/-/en/Marty-Cagan/dp/1119387507/ref=sr_1_1?crid=1R5JU9NNBPIVN&keywords=inspired+marty+cagan&qid=1695391216&sprefix=inspired+marty+cagan,aps,229&sr=8-1

Principles by Ray Dalio

Summary: This book outlines Ray Dalio’s life and work principles that have made him one of the world’s most successful investors and entrepreneurs. The book is designed to help people apply these principles in their own lives, providing a structured framework for decision-making and achieving goals. Great book for people that want to develop their managerial skills.

Principles: Life and Work

Principles: Life and Work : Dalio, Ray: Amazon.de: Books

https://www.amazon.de/-/en/dp/1501124021/ref=sr_1_2?crid=H0Y7XRB3GVOD&keywords=principles+ray+dalio&qid=1688420693&sprefix=principles%2Caps%2C1177&sr=8-2&_encoding=UTF8&tag=kgiamalis-21&linkCode=ur2&linkId=9560d9de8ab262df6d4538f8d57a2240&camp=1638&creative=6742

These two books serve as fundamental guides for both my managerial and product-thinking styles. While I don’t strictly follow their advice, the combination of fundamental knowledge of the books and the fresh insights I gain each year, help me continually grow and improve as a professional.

Global maxima

Thu, 21 Sep 2023 00:00:00 GMT

This approach is something I learned from reading Avinash Kaushik’s blog. It’s a simple but powerful concept on how someone should approach a solution. The idea actually comes from the world of math.

💡

Global maximum (or global maxima in plural form) refers to the highest point or value in a set or range. This is in contrast to a local maximum, which might be the highest point in a specific section of a range but not the overall highest.

Let’s look at some examples. These are problems or requests that are like local maxima and the solutions that act as global maxima.

Local maxima	Global maxima
A stakeholder request about a change in a product feature.	Develop a proper user experience that incorporated the request of the stakeholder while ensures optimal UX.
A user complaint about lack of rewards.	Develop a loyalty system that provide rewards to loyal users by incorporating 3rd party partnerships (that subsidize the cost of reward in exchange of promo campaigns).
User complaints that can’t find what he wants easily.	Develop personalized mechanisms to provide relevant options that increase the generated revenue too.
A GDPR guidance.	Develop a solution that is GDPR compliant but does not affect (or affect a little) the existing user flow & experience.
An analysis of how to increase KPI X by 5% in feature .	An analysis if feature A provides incremental revenue and how much.

Top product managers (PMs) aim for global maxima. It’s tougher, as you can see, but it’s the only way to give your customers a complete and well-thought-out experience. It’s not just about the product. You have to work well with the Marketing, Sales, and Operations teams too. They help you reach the global maximum for your project.

A good PM can make a great product, but a great PM can do that and also get everyone in the company to work together to make the product even better.

Moreover, when we talk about analysis, focusing on global maxima is crucial. Most of the time, analysts look at small changes in features:

How can we get 5% more people to use feature A?
Why did feature B lose 2% engagement yesterday?
Why did transactions go up by 3% last week?

Global maxima is about the big picture. They’re not useful for making small changes but are great for understanding if a feature is actually helping the business:

Does feature A make us more money?
What happens if we get rid of feature B?
Are we gaining loyal users?

Many PMs know their numbers, but few understand if their product is really going in the right direction.

[Step by step] How to build a chatbot using openAI, Langchain & Streamlit.

Thu, 21 Sep 2023 00:00:00 GMT

I decided to try to build a chatbot that will answer questions based on the content of my blog posts.

In order to do that I used:

Langchain, for components to load data, process text, embeddings, and conversational modeling
OpenAI for embeddings and the chatmodel,
Streamlit to publish the app.

Below is a step-by-step guide detailing the process, code, and explanations of my development journey. (PS: I used VSCode)

⚠️ Important note:

Langchain updated its library, I updated it in the article as of today (23/11/24) but I didn’t check for any other changes. The app is currently live & functional.

Install packages

First, create a requirements.txt file to specify the necessary packages. Name the file “requirements” and include the following:

langchain
openai
faiss-cpu
tiktoken
langchain-community

These packages are essential for building the chatbot. Streamlit will automatically detect and install these packages from the file. To run the program locally, you need to run the following commands in the command prompt (WIndows):

pip install langchain
pip install openai
pip install faiss-cpu
pip install tiktoken
pip install langchain-community

Import necessary libraries

Next you need to open your main file (streamlitapp.py) and install the necessary libraries. You need:

Libraries explained:

import streamlit as st:

Streamlit is an open-source Python library used to create web applications for data projects with minimal effort. It’s especially popular among data scientists and engineers for building data dashboards, visualization tools, and interactive reports.

By importing it as st, you can use the shorthand st to access Streamlit’s functions and methods.

from langchain.document_loaders import CSVLoader:

LangChain is an open source framework in Python that is widely used to develop LLM applications.

CSVLoader is a document loader in LangChain to load documents from a CSV format.

📖 There are a lot of document loaders in Langchain, you can find them here.

from langchain.text_splitter import CharacterTextSplitter:

CharacterTextSplitter is a document transformer, a text splitter in LangChain used to split text into chunks based on characters. This is useful for preparing data for embeddings.

📖 Hereyou can find other splitters as well that can be utilized with Langchain.

from langchain.embeddings import OpenAIEmbeddings:

OpenAIEmbeddings is an embedding method in LangChain designed to generate embeddings using OpenAI. Embeddings transform textual data into numerical vectors that machines can understand.

📖Find more info about OpenAI embeddings here, it’s a very popular method.

from langchain.chat_models import ChatOpenAI:

ChatOpenAI is a chat model utility in LangChain that utilizes OpenAI’s models for generating conversational responses.

from langchain.vectorstores import FAISS:

FAISS is a library developed by Meta AIfor efficient similarity search and clustering of dense vectors. Within the context of LangChain, it is used to store and retrieve embeddings efficiently.

📖Check the vector stores that support Langchain here.

from langchain.memory import ConversationBufferMemory:

ConversationBufferMemory in LangChain provides memory for conversations, ensuring the chatbot can reference previous parts of a conversation.

📖Here is more information about memory types in Langchain.

from langchain.chains import ConversationalRetrievalChain:

ConversationalRetrievalChain in LangChain combines conversation and retrieval functionalities, probably allowing the model to search through data and generate conversational responses.

📖Chains is a fundamental concept in Langchain, you can find more information here.

import os:

os is a built-in Python module that provides a way of using operating system-dependent functionality, such as reading or writing to the filesystem, managing paths, and accessing environment variables.

(Optional: How to generate the CSV)

The CSV I used, include all the text, titles and content of the blog posts I wrote so far in kgiamalis.co. In order to get them, I used the following libraries:

from langchain.document_loaders import AsyncChromiumLoader
from bs4 import BeautifulSoup
from langchain.document_transformers import BeautifulSoupTransformer
import csv

For those that are interested, I used the AsyncChromiumLoader on my blog’s sitemap to parse and load all the content from the URLs that include “blog” in their url.

Then I used beautiful BeautifulSoupTransformer, to extract h1 (titles), page_content (copy) and URL.

Load the data

CSVLoader Initialization:
Loading the Data:

#Load Data with LangChain CSVLoader
loaders=CSVLoader('personal_posts.csv', encoding='utf-8')
docs=loaders.load()

Set OpenAI key

I added the OpenAI key, which is necessary for OpenAI chat models to operate. Here you can see the way I added them within Streamlit to deploy my app.

#Set OpenAI API Key
openai_key = st.secrets["openai"]["openai_api_key"]
os.environ["OPENAI_API_KEY"] = st.secrets["openai"]["openai_api_key"]

Split content into chunks

The get_text_chunks function prepares textual data for embedding by segmenting it into smaller, more manageable chunks, ensuring both efficiency and preservation of context in subsequent processing and analysis steps.

Function Definition:
CharacterTextSplitter Initialization:
Splitting the Documents:
Return Statement:

#Prepare data for embedding
def get_text_chunks(docs):
    text_splitter=CharacterTextSplitter(separator="\n", chunk_size=1000, chunk_overlap=200, length_function=len)
    text_chunks=text_splitter.split_documents(docs)
    return text_chunks

Embed the data into a database

While there are several libraries available for storing vectors, such as Pinecone and ChromaDB, I chose FAISS for the following reasons:

Efficiency: FAISS is optimized for memory usage and speed, making it highly efficient for similarity searches.
Ease of Use: FAISS offers a straightforward API that makes it easy to store and retrieve vectors.

The get_vector_store function takes chunks of text, transforms them into vector embeddings using OpenAI models, stores these embeddings efficiently with FAISS, and returns this stored structure. This process allows for efficient similarity searches on the embedded data.

Function Definition:
Initializing OpenAI Embeddings:
Embedding the Text and Storing in FAISS:
Return the Vector Store:

#Embed the data in FAISS
def get_vector_store(text_chunks):
    embeddings=OpenAIEmbeddings()
    vectorstore=FAISS.from_documents(text_chunks, embeddings)
    return vectorstore

Create the conversation chain

This function showcases the interplay between embeddings (vectorstore), a chat model (llm), and memory management to create a conversational agent that’s both informed by past interactions and capable of retrieving relevant information from a dataset.

The Function: get_conversation_chain(vectorstore)
Language Model Initialization: llm=ChatOpenAI(temperature=0.0)
Memory Management: memory=ConversationBufferMemory(memory_key='chat_history', return_messages=True)
Creating the Conversation Chain: conversation_chain=ConversationalRetrievalChain.from_llm(...)

#Create a Conversation Chain
def get_conversation_chain(vectorstore):
    llm=ChatOpenAI(temperature=0.0)
    memory=ConversationBufferMemory(memory_key='chat_history', return_messages=True)
    conversation_chain=ConversationalRetrievalChain.from_llm(llm=llm, retriever=vectorstore.as_retriever(), memory=memory)
    return conversation_chain

Handle the user input

This function is responsible for managing the interaction between a user and a chatbot within a Streamlit application. It takes the user’s input, processes it through an active chatbot conversation, stores the chat history, and then displays the ongoing conversation in a formatted manner. If the chatbot isn’t active, it alerts the user to start the conversation.

Function Definition:
Checking for an Active Conversation:
Getting the Chatbot’s Response:
Storing Chat History:
Displaying the Conversation:
Warning for No Active Conversation:

Main function

The main function serves as the backbone of a Streamlit-based chatbot application. It sets up the user interface, initializes session variables, handles user input, and manages the chatbot’s backend processes. When a user interacts with the application, they can input their questions, start the chatbot’s processing capabilities, and receive responses, all in an interactive and user-friendly environment.

Function Definition:
Page Configuration:
Styling the Application:
Page Header:
Initializing Session State:
User Input Handling:
Sidebar Configuration:
Starting the Chatbot:
Main Execution Point:

Create the HTML template

Use this code to create your html template. Save it in the same folder with your streamlit file but in a separate file like this: HTMLTemplate.py

CSS: This part styles the chat messages. It looks like you’ve set up different styles for user and bot messages, which is great for making the interaction visually distinct.
Bot Template: This HTML structure is for the bot’s messages. It includes an avatar image and a message section.
User Template: Similar to the bot template, this is for the user’s messages. It also includes an avatar image and a message section.

Run it locally

Before running the code locally, make sure you’ve set up your Python environment properly, and you’ve installed all the required libraries as per your requirements.txt file.

Here’s how you can do that:

Install Required Packages:

Run the following commands in the command prompt:

pip install langchain
pip install openai
pip install faiss-cpu
pip install tiktoken

Set Up Environment Variables:

You might have sensitive information like API keys. It’s good to keep them in environment variables. Streamlit offers st.secrets to manage secrets, but when running locally, you may use your system’s environment variables.

setx OPENAI_API_KEY "your-openai-api-key-here"

Run Streamlit App:

Navigate to the directory where your Streamlit script is located (your_script.py), and run:

streamlit run your_script.py

A new tab should automatically open in your web browser displaying the Streamlit app for you to interact with.

Upload it on Github

There is a way to do it via VS Code as well, but here is the simplest method:

Navigate to GitHub website
Create account if you don’t have
Click on “New Repository”
Fill in the repository name and description
Choose to make it public or private
Click “Create Repository”
Upload all the files to the repository

Push it on streamlit

Again, there is a way to do it via VS Code, but here is the simplest method:

Go to https://streamlit.io/
Create account
Click on “new app”
Fill the required information

Add your secret keys by clicking “Manage app → Menu → Settings → Secrets →

[openai]
openai_api_key = "add-your-api-key"

Let streamlit run it.
You’re live. You can check the logs, from the “manage app” on the bottom right in your screen.

Here is my repository:

GitHub - kgiamalis/Chatbot-For-Own-Website

Contribute to kgiamalis/Chatbot-For-Own-Website development by creating an account on GitHub.

https://github.com/kgiamalis/Chatbot-For-Own-Website/tree/main

Here is my chatbot:

kgiamalis-chatbot.streamlit.app

https://kgiamalis-chatbot.streamlit.app/

What’s your shape?

Thu, 21 Sep 2023 00:00:00 GMT

You need to figure out which skills you want to improve and what kind of product manager (PM) you want to be. Regardless if you want to be a generalist or an expert, you must have a particular set of skills to succeed. The skills you need depend on what you enjoy doing, the companies you want to work for, and current industry trends.

Choose your skills based on:

What your company needs,
Your own likes and dislikes (especially for the long term game),
Skills you already have,
Trends in the industry,
Available training and mentors,
A PM you look up to as a role model,
Or a combination of these factors.

The shape of your product will determine the skills you develop and the kind of PM you become. Although your shape is not set in stone, it’s essential to craft it and invest effort & time to master your desired skills.

If you’re unsure where to start, you can take the free Principles assessment by Ray Dalio. It is not directly correlated with Product shape but it will help you understand yourself better:

PrinciplesYou From Ray Dalio

Find out what you're really like and how we relate at PrinciplesYou.com

https://principlesyou.com/

One great Product shape example is a concept that was coined by Shreyas Doshi and its called “10-30-50”. Here is Shreya’s tweet about the concept:

View post on X

[Review] Langchain with Python bootcamp

Sat, 16 Sep 2023 00:00:00 GMT

Today, I completed my first bootcamp on Langchain. It is the most comprehensive course I have taken so far on AI application development using Langchain. I’ve started more than five such courses, but this was the only one I managed to finish, owing to its well-organized structure and valuable content. While it is theoretical in nature, as it doesn’t delve into the actual app development process, it offers deep insights into the entire scope of Langchain.

Why Langchain?

LangChain is a modular framework for Python and JavaScript that simplifies the development of applications that are powered by generative AI language models. It provides a number of features that make it a valuable tool for AI application development, including:

Ease of use: LangChain is designed to be easy to use, even for developers with no prior experience with LLMs. The framework provides a high-level API that abstracts away the complexity of working with LLMs, so developers can focus on building their applications.
Flexibility: LangChain is a flexible framework that can be used to build a wide variety of AI applications. The framework supports a variety of LLMs, and it can be used to create applications that perform a variety of tasks, such as generating text, translating languages, and writing different kinds of creative content.
Scalability: LangChain is designed to be scalable, so it can be used to build large and complex AI applications. The framework supports distributed training and inference, so it can be used to train and deploy LLMs on large datasets.
Community support: LangChain has a large and active community of users and developers. This community provides support and resources for LangChain users, and it helps to ensure that the framework is constantly evolving and improving.

Overall, LangChain is a powerful and versatile framework that can be used to develop a wide variety of AI applications. It is a valuable tool for developers who want to build applications that are powered by LLMs.

Pros of bootcamp:

Provides detailed sections on crucial aspects of Langchain development.
Features collab notebooks with functional code for every section.

Cons of bootcamp:

Does not cover the development of actual LLM apps.
Lacks methods on evaluating the quality of your LLM application.
While the code in collaborative notebooks is functional, it would be beneficial to include more comments.

Course URL:

LangChain with Python Bootcamp

Build real world applications with Large Language Models and LangChain!

https://www.udemy.com/course/langchain-with-python-bootcamp

Topics covered:

Models - Input & outputs
Data connections
Chains
Memory
Agents

Lessons on advice

Fri, 15 Sep 2023 00:00:00 GMT

Today’s lesson is brief and to the point.

On giving advice

You can offer many types of advice, such as:

Advice from your own experiences,
Advice from what you’ve learned in books,
Advice from others’ mistakes,
Advice based on your instincts about a situation, etc.

However, there’s one type of advice I suggest you avoid giving:

Advice that nobody asked for.

Avoid giving unsolicited advice. It may seem obvious, but you’ll often find yourself advising people who haven’t asked for it. You’ll also receive advice you didn’t request. While I don’t believe it’s usually done with bad intentions, unneeded advice is often ignored. Also, giving unsolicited advice can make you appear intrusive.

A relevant quote from Stoic philosophy:

You always own the option of having no opinion. There is never any need to get worked up or to trouble your soul about things you can’t control. These things are not asking to be judged by you. Leave them alone.

~ Marcus Aurelius

Focus on sharing advice with those who are open to it, and remember to apply it to your own life too.

On receiving advice

Another fascinating aspect to consider is whose advice warrants serious consideration. In our lives, we often encounter unsolicited advice, but it’s crucial to discern which pieces of guidance truly hold value. A principle I find particularly insightful in this context is Ray Dalio’s concept of “believable opinions,” which can be aptly adapted to the realm of advice.

Consider this: the most credible advice typically comes from individuals who have not only achieved success in the specific area they’re advising on but also possess the ability to logically articulate the cause-and-effect dynamics of their insights. This concept of evaluating the credibility of advice based on expertise and logical reasoning is instrumental. When we apply this filter of believability to the advice we receive, our decision-making process becomes not only more effective and equitable but also more aligned with our goals and values.

Even when we might disagree with a particular piece of advice, understanding its foundation in experience and logical reasoning allows us to respect and potentially stand behind the decision it informs. This approach to considering advice ensures that we’re not swayed by every opinion that comes our way, but rather, we give weight to those insights that are grounded in proven success and thoughtful analysis.

Here’s how Ray Dalio encapsulates this idea:

“The most believable opinions are those of people who 1) have repeatedly and successfully accomplished the thing in question, and 2) have demonstrated that they can logically explain the cause-effect relationships behind their conclusions. When believability weighting is done correctly and consistently, it is the fairest and the most effective decision-making system. It not only produces the best outcomes but also preserves alignment, since even people who disagree with the decision will be able to get behind it.”

This philosophy transforms how we perceive and value the advice we receive, encouraging us to seek guidance that is not only sound but also substantiated by real-world success and logical clarity.

The curse of knowledge

Wed, 06 Sep 2023 00:00:00 GMT

In my daily work, I collaborate with various professionals: from finance experts and accountants to software developers, designers, and marketing specialists. Each is a master in their respective domains.

While product managers are also specialists, the realm of Product thrives on a “generalist” ethos. Today, I want to shed light on a prevalent misconception often observed when interacting with experts: the “Curse of Knowledge”. I first recognized this in myself when attempting to describe Product management to an individual unfamiliar with the sector.

The “Curse of Knowledge” is a cognitive bias. It arises when individuals deeply versed in a topic struggle to see it from the perspective of those less informed. Essentially, once you’ve internalized a piece of knowledge, it becomes challenging to recall not knowing it.

This cognitive situation can obstruct clear communication, project explanations, and alignment. When a knowledgeable person presumes others possess the same background understanding, they risk creating confusion or a sense of exclusion.

The Famous Experiment from “Made to Stick”

In “Made to Stick”, authors Chip and Dan Heath reference a study led by Elizabeth Newton in 1990. Participants were categorized as “tappers” or “listeners”. Tappers would relay recognizable tunes, such as “Happy Birthday”, using table taps, while listeners tried to identify the melodies. Although tappers anticipated a 50% success rate for the listeners, a mere 2.5% identified the songs. This experiment vividly illustrates the Curse of Knowledge: the tappers found it difficult to dissociate their comprehensive awareness of the song from the rudimentary taps.

(such a great example to explain curse of knowledge)

How to Spot It:

Assumption of Understanding: If you find yourself thinking, “It’s obvious” or “Everyone knows that”, you might be under the curse. CVR does not mean conversion rate for everyone.
Frustration: If you’re getting impatient with someone not grasping what you’re sharing, it’s a sign.
Overuse of Jargon: Using specialized language that only experts would understand.

How to Deal with It:

Simplify: Break down complex ideas into basic concepts. Avoid jargon.
Ask for Feedback: Encourage your audience to ask questions or summarize what they’ve understood.
Use Analogies and Stories: These can bridge the gap between what you know and what your audience knows.
Practice Active Listening: Ensure you’re paying attention to verbal and non-verbal feedback.

In Product management, effective communication is more than important. Recognizing and mitigating the Curse of Knowledge ensures that both “tappers” and “listeners” share mutual understanding.

[Guide] Anomaly detection algorithm

Sat, 02 Sep 2023 00:00:00 GMT

Anomaly detection is a data-driven concept that is widely used by businesses across industries in order to identify potential anomalies in a performance of a product. The concept applies in multiple situations, from fraud detection, to performance monitoring, budget optimization etc.

For this demonstration, we’ll be using a simple dataset and the Prophet algorithm. Its not a pure anomaly detection algorithm but it can serve this purpose too. We’ll work with a dataset detailing prices for various fuel types. You can download it from this website (from the 1st table).

We’ll be using a Collab notebook, so there’s no need to download any additional tools to run the provided code.

#Download necessary libraries
!pip install pandas
!pip install matplotlib
!pip install prophet

#Load necessary libraries
from prophet import Prophet

# Initialize the model and set its sensitivity
model = Prophet(interval_width=0.95)

# Fit the model
model.fit(data)

# Forecast on the original data to get the bounds
forecast = model.predict(data)

#Calculate the anomalies plus the upper and lower bounds.
anomalies = data.loc[(data['y'] > forecast['yhat_upper']) | (data['y'] < forecast['yhat_lower'])]

#Visualize the results
import matplotlib.pyplot as plt

# Plot the Prophet forecast
fig1 = model.plot(forecast)

# Overlay the anomalies
plt.scatter(anomalies['ds'], anomalies['y'], color='red', s=50, label='Anomalies')
plt.legend()
plt.show()

#The red dotes are dates that are considered as anomalies.

#Print the data points that were flagged as anomalies
print(anomalies[['ds', 'y']])

							ds      y
1806  2022-02-22  1.621
1807  2022-02-23  1.623
1808  2022-02-24  1.626
1809  2022-02-25  1.635
1810  2022-02-26  1.640
...          ...    ...
1932  2022-06-28  2.123
1933  2022-06-29  2.120
1934  2022-06-30  2.112
1935  2022-07-01  2.102
1936  2022-07-02  2.094

Run the code on your own, adjust the interval_width and check the different results that will be generated. Also, try to expand the capabilities of Prophet or try other algorithms (like Luminaire) to get a better understanding of how anomaly detection works.

Awareness test

Fri, 01 Sep 2023 00:00:00 GMT

Here is a 15-year-old (but useful) test about awareness.

I don’t want to spoil it for you, so please check the video first and then read the rest of the article.

Video:

https://www.youtube.com/watch?v=oSQJP40PcGI

The reason I find this video very helpful is that, especially when it comes to analytical thinking & decision making, there are many biases. Even after creating a detailed analysis or getting a sophisticated decision, we might still be blind and miss the obvious things.

I’ve shown this test to many people, yet only a few managed to spot the bear. Almost everyone managed to count the passes correctly, though.

There are many times I see analyses and features that seem right but still didn’t manage to answer important questions.

Similarly, we might check a ton of parameters of a project but we managed to miss some of the most important ones.

Why is that happening?

Here are a couple of reasons:

Complexity

One of the most popular complex examples in the business world is how to measure incrementality. Your feature might generate thousands of interactions and conversions, but are those conversions incremental? A simple question with a complex answer.

There are many ways to measure incrementality. A common one is controlled experiments & holdout groups, but they require an initial setup. If you miss that setup, you can’t use them.

If you missed setting up controlled experiments, causal impact is a statistical method developed by Google. It can be useful to measure incrementality, from product features to marketing campaigns & sales efforts.

Sunk cost

Imagine that you spent a significant amount of time building a feature, or you used a large portion of your marketing budget on a campaign. Your investment isn’t only resource-based; it’s emotional too.

Unconsciously, you’ll try to find ways and metrics to prove to yourself and others that this initiative worked. In cases of great success or failure, things are straightforward, but most initiatives are not outliers.

Confirmation bias

A common bias for everyone. As humans, we tend to seek, interpret, and remember information in a way that confirms our pre-existing beliefs or hypotheses. Even if you’re data-driven, the vast array of available metrics can be misleading.

You might cherry-pick those that confirm your bias and make a decision that feels data-driven, even if it’s not. There’s a relevant quote from British economist Ronald Coase: “If you torture the data long enough, it will confess.”

Expertise

Your expertise can sometimes make you blind, as you might focus only on what you know and inadvertently overlook other crucial aspects that fall outside your scope. A prime example of this is when one evaluates the success of a feature while completely neglecting user incentive campaigns.

This oversight can happen simply because you might not be aware of the active campaigns, especially if campaigns are handled by the Marketing team, while product features are developed by the Product team.

Title

Lastly, the organizational hierarchy plays a role. When a CEO or C-level executive makes a request, it carries more weight than a regular stakeholder’s request. Our attention tends to shift towards expedient delivery (e.g., “as soon as possible” or “needed yesterday”) rather than a comprehensive review and analysis of the situation.

Resulting

Tue, 29 Aug 2023 00:00:00 GMT

Football teaches us a lot, even for those not on the field. Here’s a story about two famous names in football: Mourinho and Balotelli. Here’s how Mourinho tells the story:

Let’s dissect a tale involving Mourinho and Balotelli, two towering figures in the world of football. Here’s the story recounted by Mourinho himself:

“I remember one in Kazan. We went to Kazan in the Champions League. In that match all my strikers were injured. No Milito, no Eto’o, I was really in trouble and Mario was the only one.

“Mario gets a yellow card in minute 42nd, 43rd. So when I go to the dressing room at half-time, I spend 14 minutes of the 15 speaking only for Mario.

“Mario I cannot change you. I cannot make a change. I don’t have a striker on the bench. Don’t touch anybody. Play only with the ball. When we lose the ball, no reaction. If somebody provokes you, no reaction. If the referee makes a mistake, no reaction. Mario, please.

But, Mario, who was only 19 years old then, got another yellow card and was removed from the game in the 60th minute.

What’s the lesson?

Mourinho knew Mario might get into trouble, so he used the whole break to talk only to him. It was the right call.

But even though Mourinho made a good decision, things didn’t go well. So, what happens when you do the right thing, but it still goes wrong?

This is not rare. There is a fallacy called “resulting” which is dangerous since it prevents us from properly evaluating whether we’re making good decisions, because we tie our decisions solely with their outcomes.

Sometimes we win or lose by factors we can’t control, regardless of the decision we made.

For these reasons, we need to understand that our end results alone don’t tell us the whole story, whether we’re good making good decisions.

Here is a chart from fs blog the describes combinations between outcomes and process behind them.

As you can see, a flawed process often relies heavily on luck. You might strike it lucky and achieve a positive outcome, or you might end up attributing negative results to bad luck.

The situation where one follows a “good process” but still ends up with a “bad outcome” is especially challenging to handle.

My two cents: Consider investing time in optimizing your decision-making process by delving into the origins of your decisions and their subsequent results. Over the past year, I’ve made a concerted effort to document my decisions, along with the reasoning behind each one, reviewing the process once outcomes are realized. This practice, known as “journaling,” has proven effective, although it can be challenging to maintain consistently.

If you’re interested in refining your decision-making skills further, consider reading some books on the subject. If you’re unsure where to begin, here are a few recommendations:

Thinking in bets by Annie Duke

Thinking in Bets: Making Smarter Decisions When You Don't Have All the Facts

Thinking in Bets: Making Smarter Decisions When You Don't Have All the Facts : Duke, Annie: Amazon.de: Books

https://www.amazon.de/-/en/Annie-Duke/dp/0735216371/ref=sr_1_3?crid=LK8NZ84CL8Y5&keywords=thinking+in+bets&qid=1696352404&sprefix=thinking+in+bets%2Caps%2C129&sr=8-3&_encoding=UTF8&tag=kgiamalis-21&linkCode=ur2&linkId=c74bd1922f47dafa02c66e97b9fb49e0&camp=1638&creative=6742

Decisive by Chip & Dan Heath

Decisive: How to Make Better Decisions

Decisive: How to Make Better Decisions : Heath, Chip, Heath, Dan: Amazon.de: Books

https://www.amazon.de/-/en/Chip-Heath/dp/1847940862/ref=sr_1_1?crid=ACPFR8ZC3JI&keywords=decisive+by+chip+and+dan+heath&qid=1696352458&sprefix=decisive%2Caps%2C125&sr=8-1&_encoding=UTF8&tag=kgiamalis-21&linkCode=ur2&linkId=aa9fb1eca3a284340f5dfc7e80dcc809&camp=1638&creative=6742

The Tail End theory

Mon, 28 Aug 2023 00:00:00 GMT

Actually this is not a productivity article but it is still useful. A few years ago, I was part of the Marketing team. Now, I find myself in the Product department. From my perspective, I’ve had the opportunity to work with some of the most limited resources in a company:

Technical capacity
Marketing budget

However, this post touches on what I believe to be the most significant constraint: Time.

There is an intriguing theory named “The Tail End” which dives deep into this very concept. I came across a fascinating article about it on a blog called “Wait But Why” (I’ll share the link at the end).

Here’s a brief overview of “The Tail End”:

Imagine you’re 35 years old, just like me.
• Your father is 60, as is mine.

Having relocated to another city or perhaps even a different country, your meetings with him are limited. You manage to spend only around 30 days with him each year. These are distributed as 8 days during Christmas, 8 days over Easter, 8 days in summer, and the remaining days scattered throughout the year for occasions like birthdays or adhoc visits.

The average life expectancy in Greece for men stands at 79.8 years. To make calculations easier let’s make it 80.

By this logic, you’d have 600 days left (30 days annually plus the 20 years he has from the age of 60 to 80) with your father. And it’s essential to remember, once a day goes by, it doesn’t return.

600 days are less than 2 years. We’re in the tail end.

The article shared some startling insights:

By the time we finish high school, we’ve already used up about 93% of our face-to-face time with our father. Presently, we’re relishing the last 7%. We’re in the tail end.

It might come across as slightly grim, but it’s an undeniable truth. The central message of the article is the following:

Life’s brevity means that setting priorities and valuing quality time is of utmost importance.

You can read it here:

The Tail End — Wait But Why

No matter what your age, you may, without realizing it, be enjoying the very last chapter of the relationships that matter most to you. Make it count.

https://waitbutwhy.com/2015/12/the-tail-end.html

[Review] LangChain for LLM Application Development

Sun, 27 Aug 2023 00:00:00 GMT

Today, I completed a concise course on Langchain offered by deeplearning.ai. In this course, Andrew Ng collaborates with Harrison Chase, not the creator of Langchain, but an expert in the field.

This is my second course on LLMs. Despite its brevity, the course offers a comprehensive overview of Langchain. Even though the course duration is listed as one hour, I invested over five hours due to taking notes, running my own examples, and some minor debugging.

The course delves into topics such as:

Models, prompts & parsers.
Memory handling.
Chains.
Question & answer.
Evaluation (with an Easter egg).
Agents.

The course emphasizes hands-on learning, providing a collaborative notebook for each section. Every segment is practical, packed with relevant examples, ensuring learners grasp how Langchain operates in real-life scenarios.

I would wholeheartedly recommend this course to anyone keen on developing applications powered by LLMs.

Link:

LangChain for LLM Application Development - DeepLearning.AI

The framework to take LLMs out of the box. Learn to use LangChain to call LLMs into new environments, and use memories, chains, and agents to take on new and complex tasks.

https://www.deeplearning.ai/short-courses/langchain-for-llm-application-development/

An interesting retention campaign

Wed, 23 Aug 2023 00:00:00 GMT

Back when I was working in Marketing, we were on the hunt for effective ideas for retention campaigns. Retention campaigns target users who haven’t placed an order for many months, with the aim of persuading them to return or at least inform us why they abandoned our service (so that we can improve).

I’ve encountered a wide variety of campaigns, ranging from the fancy and creative to those offering substantial incentives, but none have stood out quite like the following campaign. It was crafted by the Marketing team at Yemeksepeti. No need to describe or highlight why its great, just watch it and you will understand.

Suç ve Ceza (English Subtitles) - Yemeksepeti

Yemeksepeti sosyal medya hesapları: Yemeksepeti Twitter: http://ymk.st/yemeksepetitwitter Yemeksepeti Facebook: http://ymk.st/yemeksepetifacebook Yemeksepeti Instagram: http://ymk.st/yemeksepetiinstagram Dünyanın en büyük mutfağı Yemeksepeti’nin YouTube kanalına hoş geldiniz. Yemek kültürüne dair Türkiye’den ve dünyadan ilham alarak oluşturduğumuz içerikler kanalımızda sizleri bekliyor. Yemeksepeti ile binlerce restoran, yüz binlerce yemekten istediğinizi anında online sipariş edebilirsiniz. -- Yemeksepeti Resmi Hesabıdır. The official Yemeksepeti YouTube Channel

https://www.youtube.com/watch?v=SHtl78HAbco

Note: The campaign launched at 2017.

We need more retention campaigns like this.

Ask Great Questions

Tue, 22 Aug 2023 00:00:00 GMT

An underappreciated skill, I have yet to find a single product manager (PM) who asked great questions and wasn’t competent in their role. Asking excellent questions might seem like an unconventional way to identify a top 1% PM, but I encourage you to experiment with this idea and let me know if it works.

⚠️ Disclaimer: It’s not just about asking questions. It’s about asking good to great questions. Quality, not quantity, is what counts here.

Asking great questions implies:

Being a Good Listener: If you’ve asked an insightful question, it means you’re already an attentive and engaged listener; otherwise, your question might fall short. I want to stress that even basic listening is not a given for many professionals, so being an active listener can indeed set you apart as a top-tier PM.
Confidence: In companies rife with political maneuvering, you need the self-assurance to ask questions, particularly when meetings involve high-level executives. The most competent PMs have the confidence to ask what they need to know, phrasing it the right way to guide the conversation in the desired direction. A great question doesn’t have to be complex; it might be straightforward or even seem “silly” initially.
Building the Right Thing vs. Being Right: While PMs may have differing opinions, they ask questions aimed at guiding the discussion towards constructing the correct solution, rather than merely asserting their correctness or forcing their viewpoint.
Impact-Driven Questions: The best PMs don’t ask questions just for the sake of it. They’re not trying to impress others with their knowledge, make others feel uneasy, cultivate politics, or establish themselves as clever. They ask questions to propel the conversation forward and maintain a focus on the impact.

Understanding this principle can be instrumental in recognizing and nurturing top product management talent. Asking great questions, while seemingly simple, can reveal a deep understanding of what truly matters in the product development process, and reflects attributes that set the most successful PMs apart from the rest.

User acquisition KPI #2

Mon, 21 Aug 2023 00:00:00 GMT

Continuing the series on user acquisition KPIs, I want to introduce a metric that might not be widely recognized but can serve as a valuable health indicator. This calculated metric offers insight into the experience of a new user and, in our particular situation, could reveal underlying product or business issues.

KPI:

Percentage of user acquisitions that signed up and converted within the same day, referred to as “single touch acquisitions.”

Definition:

New users who registered and placed their first order on our platform within the same day.

Why it matters:

The online food delivery marketplace is an industry that typically is high-intent and it involves immediate decisions. Most of the time, users who land on our platform are hungry and eager to order. Industries that cater to urgent needs often generate high user conversion rates, so we’ve found that the proportion of people who sign up and order within the same day is a valuable health indicator.

A low percentage of single touch acquisitions could signal a poor user experience. This might be due to a variety of reasons, such as technical downtime or supply issues like closed shops, limited offers, or marketing campaigns promoting something that’s already expired.

It’s different to generate 1,000 acquisitions with 1,200 signups compared to generating 1,000 acquisitions with 4,000 signups. The end result may be the same, but the share of single touch acquisitions varies significantly.

Challenges:

Some teams promote and even incentivize signups on their platforms, although the reason for doing so might not always be clear. A marketing campaign focused on this can negatively impact this KPI.

This metric emphasizes a specific time window that might not apply to every business. For example, if you run a leasing company, this KPI is unlikely to be relevant (unless you work in instacar 😉). Additionally, a user’s order might not necessarily occur within the same session or on the same device as the signup, adding complexity to this measurement.

User acquisition KPI #1

Fri, 18 Aug 2023 00:00:00 GMT

In this series of posts, I’ll share some key KPIs (key performance indicators) about user acquisitions. As we already know, not all customers are created equal, so we’ve invested a significant amount of time and effort to identify the right KPIs. These indicators help us understand the quality of new customers and guide our strategy and budget towards optimal performance. I’ll skip the more common ones, such as device, landing page, marketing channel, 1-month cohorts, etc., to focus on those that may not be as popular but are still important.

Due to confidentiality, I won’t share our specific numbers, but I’ll offer some aggregated-level KPIs that I believe are valuable for various consumer products.

KPI:

Share of user acquisitions generated through word of mouth (WOM)

Definition:

New users who place their first order on our platform and, when asked, “how did you learn about us”, respond with, “a friend told me about your platform.”

To enable this, by the time a new user performs his first order he will see a survey “how did you learn about us” with a list of options that correlate to communication channels, in a randomized way.

Why it matters:

WOM acquisitions are pivotal for several reasons:

When a user recommends our platform to someone else, it indicates they had a positive experience. Satisfied users are the cornerstone of every thriving company. They often serve as our most effective promoters. Notably, WOM has consistently been among our top three feedback responses since our business began.
This kind of recommendation builds trust and positivity for new users. Starting with a positive outlook is crucial because it enhances the new user’s likelihood to complete an order. Some users even mentioned having friends guide them through their first order, ensuring they understand the process. It’s like having a strong onboarding system in place.
Organic referrals from existing users often come at no or minimal cost. A lower Customer Acquisition Cost (CAC) means more resources can be allocated elsewhere, such as experimentation or further growth.

Challenges:

Two main challenges come with this metric:

The metric relies on user feedback, which can be prone to type I and type II errors, especially with limited sample sizes. A user’s first order is a unique event, and considering we’ve grown significantly over the years, the proportion of new users to existing ones is smaller.
Optimizing this feedback isn’t straightforward, particularly without incentives. We can’t predict when an existing user will recommend our platform to a potential new user. While we always aim to provide the best user experience, several factors might encourage a user to recommend our service. These can range from unique offerings to excellent customer support or even habit formation. However, it’s challenging to determine which aspect most influences a user’s decision to recommend us.

[Review] AI worskhop for Senior PMs

Fri, 18 Aug 2023 00:00:00 GMT

In June, an engaging workshop took place, hosted by Marily Nika, a Greek Product leader with a wealth of experience in AI, having worked for companies like Meta and Google.

Honestly, the title and its focus on senior Product Managers caught my eye, so I decided to participate. The workshop indeed attracted many senior PMs. Although it provided a great networking opportunity, I missed out on this aspect since I didn’t watch the entire live workshop but only 50% of it (the rest 50% I watched it on my own pace). Consequently, I wasn’t engaged in the discussions that occurred within the workshop and on the Discord server.

After revisiting the educational content last week, I wanted to write a brief review about the workshop. The workshop consisted of 6 sessions, addressing a broad spectrum of AI topics. I’ll break down the learnings into key categories to cover the essential parts:

Introduction:

5 stages of AI Product management,
Evolution of LLMs,
Risk & misuses of LLMs,
ML Ops overview and stack,
AI on the edge.

Technical:

LLM technology stack,
Deploying LLMs (live),
Architecture overview,
How to deploy a robust AI system,
End-to-end observability.

Career:

Breaking into AI PM career (checklist),
AI tool stack for Product teams,
Live review of job ads,
Live review of CVs.

Overall, the workshop was fascinating, covering vital areas for aspiring AI PMs. Marily proved to be an engaging instructor who resonated well with the audience and communicated essential information effectively. Her hands-on experience in AI enriched the entire session, especially during the career segment, where she offered invaluable insights on becoming a PM, current recruiting trends, and dos and don’ts for AI PMs. There were even mock AI PM interview questions and answers!

Regarding feedback, I wished for more analysis and teardown of a few successful AI products, explaining why they succeeded and how to build a successful product, would have been enlightening. This is relevant today as numerous tools and startups focus on AI, but evaluating their market value and market fit can be challenging. Also, providing more technical sessions and examples (such as fine-tuning a model) would be really helpful for the PMs because if you understand the whole process, you will become a more competent AI PM.

Last but not least, the sessions featured valuable presentations from other respected professionals in the field. These were particularly insightful as some experts were senior AI PMs who shared valuable tips based on their experience. Additionally, there were experts who, while not PMs themselves, work closely with PMs. They provided advice on what they expect from a competent AI PM, as well as what they don’t expect. These diverse perspectives added depth and practical understanding to the content of the workshop.

The course is recommended for senior PMs that want to break into AI. In case anyone is interested in joining the next cohort of the session here is the link:

Advanced AI Product Management with Technical Deep Dive by Dr Marily Nika by Marily Nika on Maven

A 3-week deep dive into complex AI concepts, problems and challenges you will face as an AI PM and how to solve them. Learn from 5 leaders.

https://maven.com/marily-nika/advanced-ai-product-management

This course is part of the AI Product management bootcamp:

AI Product Management Bootcamp by Marily Nika on Maven

This 12 week program aims to cultivate a new generation of AI product leaders who can drive innovation in the rapidly evolving AI ecosystem

https://maven.com/marily-nika/ai-pm-bootcamp

Path to mastery

Thu, 17 Aug 2023 00:00:00 GMT

Recently, I finished reading the book “Mastery” by Robert Greene. Greene is an intriguing author, somewhat controversial, with a mix of supporters and critics. In this book, he outlines the three phases someone should follow to attain mastery—the sense of having more control over reality, others, and ourselves.

According to his theory, there are three distinct phases or levels to achieve mastery:

1st: Apprenticeship

Stand on the outside of the field, absorbing as much as you can about the basic elements and rules. The principle is simple: the goal of an apprenticeship isn’t money, a good position, or title, but rather the transformation of your mind and character. It’s the first step on the journey to mastery.

2nd: Creative-active

Through extensive practice and immersion, we gain insight into the inner workings, how things connect, and thus understand the subject more comprehensively.

3rd: Mastery

This phase involves the ability to experiment and creatively play with the elements involved. Our knowledge, experience, and focus become so profound that we can see the entire picture with complete clarity.

Besides the mastery process, there’s a dedicated chapter about mentorship, discussing its importance, finding the right mentor, and maximizing a mentorship.

Additionally, there’s a chapter about social intelligence and the ability to read people, understand their worldview, and recognize their individuality. In my view, this is the most compelling chapter of the book, encompassing strategies like:

Speak through your work,
Craft the appropriate persona,
See yourself as others see you,
Suffer fools gladly.

The book is filled with fascinating stories from influential personalities like Charles Darwin, Benjamin Franklin, Wolfgang Amadeus Mozart, and Marcel Proust, illustrating how they achieved their mastery.

Some of book’s highlights:

Natural talent or high IQ cannot explain future achievement
Our levels of desire, patience, persistence and confidence end up playing much larger role in success than sheer reasoning powers.
In moving towards mastery, you are bringing your mind closer to reality and to life itself.
To follow precisely the lead of others or advice from a book is self-defeating.
The future belongs to those who learn more skills and combine them in creative ways.
To attain mastery, you must adopt what we shall call Resistance Practice: Go to the opposite direction of all of your natural tendencies when it comes to practice.
Mentors do not give you a shortcut, but they streamline the process.

I found the book interesting for a couple of reasons:

Firstly, the process that is dealing with is an interesting open-ended framework that is close to reality. You can’t become a master without acquiring and mastering the right skills.

Moreover, I appreciated the analogies between the process and the stories of influential individuals. This connection can help readers understand how the framework aided them and how it might assist others.

Lastly, the section on social intelligence was intriguing, offering valuable insights and tips on dealing with people effectively.

I recommend this book to those beginning their professional journey, as well as experienced individuals who may be struggling to unlock their mastery phase.

PS: It comes in a concise edition too.

Mastery: Robert Greene (The Modern Machiavellian Robert Greene)

Mastery: Robert Greene (The Modern Machiavellian Robert Greene) : Greene, Robert: Amazon.de: Books

https://www.amazon.de/-/en/Robert-Greene/dp/178125091X/ref=sr_1_1?keywords=mastery+robert+greene&qid=1692438527&sprefix=mastery,aps,766&sr=8-1&_encoding=UTF8&tag=kgiamalis-21&linkCode=ur2&linkId=19a20904fdf14680870d856b5bb14bda&camp=1638&creative=6742

Don’t start as PM

Mon, 14 Aug 2023 00:00:00 GMT

I began my career in marketing, with a specific focus on performance marketing. This concentration honed my skills in several pivotal concepts, including:

CRO (conversion rate optimization)
Optimization of user funnels and mapping user journeys
User segmentation (e.g., new users, returning users, churned users)
Development of personas
Utilization of Google Analytics
Designing retention and loyalty campaigns

These concepts proved invaluable for my progression in the product domain. Fortunately, my existing expertise in these areas made it easier for me to grasp marketing challenges and proactively design superior solutions.

I’d advise against starting directly as a PM. The most accomplished PMs I’ve encountered didn’t begin their careers as PMs. Instead, many initiated their journeys as:

Engineers
Product designers
Marketers
Growth hackers
Analysts
Operational specialists

Each of these roles can pave the way to becoming an exceptional PM. They equip you with a unique perspective and specialized knowledge, both of which can significantly impact your organization.

An often-overlooked benefit from my marketing days was my role as a stakeholder. I was on the “other side”, striving to convey the significance of my requests to the Product/Tech teams. Having been a stakeholder is invaluable for a PM. It fosters a genuine understanding and empathy towards others in different departments seeking your support. Trust me, empathizing with a stakeholder’s position is entirely different when you’ve been in their shoes.