What the product job actually is

It’s been a year since I formally took on the CPO role, and about a decade building products before that, so I wanted to share a few reflections.

Product is the second most exciting job in a company, after founding it. You get nearly the CEO's exposure without the parts I'd happily skip such as fundraising, finance, HR across the whole org. What you keep is the part that's actually fun: deciding what the company builds and why.

It is also a line of work where you need to continuously balance scientific approach to solving problems with art (what the industry refers to as “taste” these days). As Ivan Zhao, Notion's founder, explained it not too long ago: put too much of yourself in and the product becomes an art project; lean too far into the business and it becomes a commodity. I feel that tension every day.

And it’s the kind of job where you make important decisions under incomplete information. Sometimes you have too much of it and the right call still isn’t obvious. More often you’re missing context but have to be decisive anyway.

You could argue this is exactly why startups outcompete big companies, acting on imperfect information compounds, but it’s also highly stressful.

You make those calls at the intersection of what’s feasible with today’s resources, what the business needs, where the industry is heading, what users are asking for, what the data shows, where leadership wants the company to go, and what you believe to be possible.

Naturally, you become opinionated. People challenge your decisions constantly, because for one thing to get built, another has to wait.

To make things worse, every input you get is likely corrupted…

Users rarely understand their problems which leads to giving you wrong signals. It’s genuinely hard to reach real insight just by talking to users, which is why Steve Jobs wasn't a fan of them.

Founders want things done fast and well, and they pivot their thinking often as the market moves and feedback pours in from across the org.

Engineers push back for security, scalability, and whatever they consider feasible.

You can’t reliably lean on data either because data reflects past behavior. So when the data disagrees with your product intuition, the first step is usually to question whether the system is measuring the full reality correctly (most often, something, somewhere, isn’t tracked well).

In turn, your job becomes integrating all that without over-fixating on any single one.

That’s the actual job. Not the glamorous product launches most expect…

Three things I had to learn the hard way:

1. You do not have the complete authority to make every call about what gets built. In reality, you have to continuously story-tell and convince every stakeholder why your approach makes sense. Smart people only do their best work when they’re convinced. One caveat, once your judgment proves effective, people grant you more freedom, ship one underperformer and suddenly product decisions need a defense, so back to storytelling!

2. A second misconception is that you just need to read the data to make good decisions. I have learned that product intuition is equally important, and it’s far harder for competitors to replicate. I argued this at length in a separate essay.

3. A third is that resources solve all of your problems. True for some companies. But many start to suffer as they scale, drifting toward entropy and bloated scope. Prioritization gets harder, not easier, because everyone now believes you can do far more (AI amplifies this further). More features don’t make users happier. Shortcuts you took years ago come back to bite you. More people means passing context becomes exponentially harder.

Which brings me to the thing I have come to believe the most. I’ve come to see the roadmap less as a plan and more as a portfolio of experiments. Most bets won’t work, a few hopefully will prove important. The skill isn’t picking winners because you can’t (at least not 100% of the time), it’s structuring the portfolio so the bets compound when they’re right and stay contained when they’re wrong. A failed experiment should cost you that experiment, not the three next to it.

That’s also why I think product is the best preparation for founding a company. In fact, I have noticed that many of my founder friends tend to step into product roles, some to catch their breath from the grueling founder journey, others to sharpen themselves before starting the next company.

AI is collapsing the cost of building. And when building is cheap, the bottleneck moves back to the only thing that was ever really scarce: deciding what’s worth building at all.

Recently I was interviewed by my friend Katie for her podcast, Beyond the Raise. A few people reached out to say they enjoyed it, so I’m sharing the conversation here: