Reassemble the forest

DEMMANND is the answer to a question. It isn't a company that went looking for one.

The question was what's actually broken about the way enterprises sell into a market where buyers decide before anyone talks to them. Not which gap to fill first. What's actually wrong.

The normal move at that point is to name a thing and go build it. Someone describes a gap, you add a feature that fills it, everybody agrees it's reasonable, and you're off. There were plenty of candidates and every one of them would have been defensible.

We stayed on the problem instead, until it was stated properly. The statement is one sentence: enterprises aren't short of intelligence, they aren't actioning the intelligence they already have.

Every candidate we could have built added intelligence. That's why none of them would have mattered.

Once the problem was stated that way, what it needed was clear, and what it needed didn't exist. That's the company.

The hard part isn't stopping

Everybody says they ask what the real problem is. Stopping isn't the hard part.

The hard part is not having to be right. Not needing the answer to be one you already had, or one that makes your last two years look good, or one you can defend in the room where you said something else. You're after what's actually true about the thing you're trying to solve. Then you let the answer fall out of that.

Most root cause work I've watched is reverse-engineered. Somebody already knows what they want to build, and the investigation goes looking for the problem that justifies it. It has all the same steps. You ask why a few times, you look at the data, you talk to customers. And you land exactly where you were always going to land, except now there's evidence attached, which makes it harder to argue with and no more correct.

What got us out of that was taking the furniture away.

What if there were no martech stack. What if there were no such thing as a website. What if agent to agent commerce isn't nearly as far off as everyone is pricing it. What if you can't actually separate information, content, and visuals, and they were the same thing the whole time.

That last one is load-bearing. The stack exists because we split them: a CMS for the words, a DAM for the images, a CDP for the data, analytics for the signal, and an enormous amount of integration work to pretend afterward that they were ever one thing. If they were never separate to begin with, most of that tooling isn't solving a problem. It's maintaining a filing decision somebody made fifteen years ago.

The agent question does the same work from the other direction. If the thing reading you is an agent, and increasingly the thing buying is too, then a website is a rendering choice rather than a place. You stop designing a destination and start keeping a record that can be asked questions.

None of that is a thought exercise for its own sake. You can't arrive at an answer that isn't a feature while the feature-shaped world is still standing, because every question you ask inside it comes back feature-shaped. That's the only shape on offer. Take the stack away and the question turns from what should we add into what is actually supposed to happen when somebody is trying to decide something, and those have very different answers.

The tell that it was honest is that what we landed on wasn't on any list we'd made. It isn't a feature, and it couldn't have been, because a feature is the only kind of answer available while you're still asking inside the existing stack. If the answer you arrive at was already one of your candidates, you probably steered.

The test

If you're choosing between solutions, you haven't found the problem yet.

A problem stated correctly doesn't hand you a shortlist. It narrows to something obvious, usually smaller than anything that was on the list before. If a team is weighing five credible options, I'd bet the problem is still described at the level of a symptom, where five things look equally good because none of them is touching a cause.

This runs against the usual advice about stepping back and seeing the big picture. Stepping back works when you already understand something and need to see its shape. When you don't understand it yet, what you get is a picture assembled out of assumptions, drawn high enough up that nothing can contradict it. That's where you get strategies that are true and useless: directionally right, operationally impossible, every sentence a category instead of a thing.

I'd rather reassemble the forest from a million trees than step back far enough that the trees stop mattering.