We were building on assumptions

February 2018

I started as product and project manager at Digital Revisor. We were building software for the accounting industry, and the founder was a state-authorised auditor with years behind him running his own practice.

He knew what he was talking about. That was the problem.

One office

We were building on the experience of one office. His own.

That experience was real enough, but it was a single data point, and we used it as though it described the whole market. Nobody did anything wrong. Everyone was competent, everyone agreed, and then nobody asks.

The phone first

I started calling around, and spoke with a good many more than I managed to meet. Five offices I went out to. All of them in Førde, and there are not many more of those here.

I was not selling anything. I just wanted to see what an ordinary Tuesday looked like for them.

The phone gave me the pattern. The visits gave me what nobody thought to mention: which systems they were already sitting in, what they did in Excel because nothing else fit, where the hours went.

It cost time

We agreed we needed more data. I was the one who went.

Weeks went into something that produced no code at all, in a small company where every hour gets logged. That was a real cost, and it was taken deliberately.

If the visits had only confirmed what we already believed, the time would have been hard to defend. They did not.

What actually changed

Two things. The solution became a different one, because I had seen where their working day stalled. And the way we talked to new customers became different, because I could talk about their day instead of our features.

A little later the same insight made the build-versus-buy call easier. We took Nextcloud as the foundation instead of building file storage, document handling and collaboration from scratch, and put our own tools on top. The product still runs on it, today as Jointly Office. The rental-tax calculator we specified in the same period is still the site's biggest traffic driver.

Not down to me alone

The founder took the objection, and we agreed this was the fastest way forward at that early stage of the build.

The designer I brought in turned the findings into something usable. The developers sat in another country and got it built through weekly standups. The security specialist built the access control.

We were not short on vision, we were short on market

The feeling was never that anyone was wrong. It was that we were building from one vision, and that we were missing the contact with the market that would eventually decide whether we were onto something.

A vision can be entirely right and still miss, if nobody has tested it against the people meant to use it. The market decides in the end, not the conviction in the room.

The test I still use

The solutions that win live inside the working day people already have. They ask nobody to move.

It is the same test I use on AI systems now. Not whether the technology is impressive, but whether anyone will actually use it on a Tuesday.