The build was finished. Then nobody logged in for twenty days.

A field service business, a system for scheduling their engineers and tracking what happens on each job. It worked. We got the team on a call, made them log in while we were watching, walked them through it. A fair bit of positive feedback, a little bit of resistance, nothing that looked like a problem. Twenty days later I checked the logins. Almost nobody had been back in.

One sentence did it

On that call, one of the technicians opened the dashboard and said, oh yeah, okay, I can see what is here. Then, half a second later: oh, what is that? Why is this there?

That was it. That was the whole thing. Everyone else in the room had been nodding, and one person asking two honest questions about two things he did not need was enough to turn the mood. It does not take much negative feedback to blow away all the positive feedback you just collected. I have watched that happen enough times now that I should have seen it coming.

And he was right. Why was it there? He is standing on a job site with a phone in his hand. What he needs is this is where I am, this is who I contact, this is what I am supposed to do, and a list he can tick off. What we had given him was a month of calendar, a set of filters, and a few things that exist because an owner asked whether they could be there.

Simpler is not the same as simple enough

The part I own is this. I had asked for it to be made simpler, more than once, over weeks. And each version genuinely was simpler than the one before it. We were all pleased with the progress. But not one of those versions crossed the line where a technician looks at the screen and says, oh, why did you not show me this before.

That is the only line that counts, and it is binary. You are either over it or you are not. Grading yourself on improvement is how you spend a month getting closer to a bar you never clear.

The instruction I gave after that call was the opposite of every instruction I had given before it. Stop making it better. Take things out. There needs to be subtraction, not addition. If one of the eight things on that screen is not used by a technician on a job site, it goes, even if it works, even if it took somebody a day to build, even if an owner likes having it.

What I do differently now

Two things, and they cost nothing.

• Watch the actual user open it cold, with no walkthrough, and count the questions they ask. Every question is a thing to remove, not a thing to explain.
• Before anything goes in, ask which person on which day uses this. If the answer is the owner might want to see it, it waits.

Both of those are just ways of forcing the same admission, which is that a feature nobody asked for is not free. It costs you the one thing you cannot buy back, which is the first ten seconds of somebody deciding whether this new system is going to help them or get in their way.

Adding things is how you show people the work you did. Taking things away is how you get the work used.