Nobody Asked Me to Write That Down

Years ago I used to take my backend developers into customer interviews.

Not only the designers and the PMs. The people who would go home afterwards and write the actual queries. We sat in a room, the customer described the problem in their own words, everyone heard the same thing at the same time, and then we went and built it. No discovery document. No research repository. The understanding lived in the room, and in the heads of the four or five people who had been sitting in it.

It worked. I still think it is the best way we ever built shared understanding, and I would do it again tomorrow.

It also left almost no trace. When someone left the company, a piece of the reasoning left with them. None of us noticed at the time, because the product kept shipping.

What We Wrote Down Was Never the Thinking

Look at what a product organization actually documents. The vision doc. The business case. The board memo. The quarterly summary. The call notes. The one-pager that goes upstairs before a decision is announced.

All of it is output. All of it is the finished shape of a decision, tidied up for an audience that was not in the room.

A model produces that in about a minute now, and produces it competently. That is the part everyone is nervous about, and I understand the nervousness. I just think it is aimed at the wrong thing. The documents were never where the value sat.

The other half never made it onto paper at all. Why we said no. What we killed and what it cost. Which evidence we actually trusted, and which we quietly stepped around because it was inconvenient. Who disagreed. Whether they turned out to be right.

I ran product for a long time and I rarely wrote any of that down.

Nobody asked me to.

On paper, a product manager produces documents. That part is free now. The part that is not free was never on paper to begin with.

A Map Is a Photo of Something That Keeps Moving

I spoke to a product leader a while back who had built a capability map in Miro. Real work behind it, three horizons, maturity stages for every capability. He had run a proper workshop. The board was good.

Two weeks later it was dead and everyone was back in Jira.

He described it to me in a way I have not been able to shake since. It was not a living organism, he said. Just a picture of one, taken on a Tuesday.

I have thought about products as living things for most of my career. A product is not a static object sitting on a roadmap. It starts somewhere small, it grows, some parts work out and others quietly do not, and occasionally a whole capability dies and you have to let it. So of course a map of it goes stale. The map was never wrong. The format was.

A separate conversation, a discovery consultant who has been doing this for years, told me his single best tool is still a physical wall covered in post-its. His biggest regret had nothing to do with methods. It was that whenever people changed positions, the documentation never carried the reasoning. The new person inherited the artifact and the why was gone. They would look at a wall, or a board, or a folder of decks, and they could see exactly what had been decided and nothing at all about why.

That is the uncomfortable version of the problem. As long as the real model of your product lives in the heads of two or three people, your product strategy has a resignation notice as its single point of failure.

The Test

Here is the thing I keep asking product leaders now. Take the last significant thing your team decided not to build.

• Can you find why, in writing?
• Can you find the evidence you based it on?
• Can you find who disagreed?
• If that person left tomorrow, does the answer survive?

Most teams go zero for four. These are good teams, by the way. That is what makes it interesting. Nobody is being careless. The work simply has no home, so it evaporates.

I watched what this costs at a company I worked at years ago. We had built a new product with partners, and early on the numbers were clearly weak. Everyone could see it. Killing it took far longer than it should have, because months of work and a proud team and a sales org that had already carried the thing into customer meetings all sat on the scale. And after we finally did kill it, sales kept it on the product list and went on selling it anyway.

That last part still amazes me. The decision existed. It just did not live anywhere that could hold it, so it did not survive contact with the next quarter.

The Work Kept Its Shape, Only the Output Got Cheap

There is a Deloitte survey that stuck with me: 3,235 senior leaders, and 84 percent had not redesigned any jobs around AI. Including the ones already using it daily.

So the shape of the work stayed exactly the same. We just made the cheap half cheaper.

If your definition of the job is producing artifacts, then yes, that job is in trouble, and the anxiety currently going around is rational. But if the job is the deciding, then almost nothing has changed except that we now have room to do the part we were always skipping.

Michael Polanyi sat with Alan Turing in Manchester in 1949 and argued that thinking runs on judgment no machine can hold. Seventeen years later he wrote the line he is remembered for, that we can know more than we can tell. He meant some of it can never be told, and fine, I believe that. But we did not even keep the part we could have told easily. We had the time. We just spent it formatting.

Give the Deciding Somewhere to Live

The fix is not another template, and it is not writing more. Teams already drown in documents.

It is treating the decision itself as the thing you keep. Not the deck that announced it. The decision, with the reasoning attached, connected to the evidence that produced it and to the work that came out of it. Something that moves when the product moves, instead of a snapshot that starts decaying the moment the workshop ends.

Sharpening your product sense is good advice and I have no argument with it. But sense with nowhere to land is just a nicer way to make the same mistake with more conviction. The best product people I have worked with still made bad calls when the deciding happened in a hallway and the reasoning stayed in one person's head.

So here is the question I would sit with, if I were still running a product org.

What is the oldest artifact in your company that still tells the truth?

If you have one, someone has been quietly doing this work for years and you should find out who. If you do not, that is not a documentation problem. That is your product strategy telling you it has never been written down at all.