The first thing I got badly wrong: I built exactly what they asked for

A client listed eleven things they wanted. I built eleven things. I was proud of it.

Two ever got used regularly. The rest sat there adding menu items and confusion, making the product feel heavier than the problem it replaced.

I had treated the requirements list like an order at a restaurant. It is not. It is a wish list assembled by several people at different moments, some guessing, at least one protecting a way of working they liked.

The subtraction conversation

The fix was not technical. It was a conversation I had been too nervous to have, because saying "you do not need that" to someone paying you feels like turning down money.

It is the opposite. Every unused feature costs the client twice — once to build, and then forever in the confusion it adds. Four things that all get used beats fourteen where nobody can find anything.

If we shipped only three of these, which three would you actually miss on the Monday?

That question is in every scoping call I run now. It is uncomfortable for about ten seconds and then people relax, because most of them knew the list was too long and were waiting for permission to say so.

What I do differently

Phase one is deliberately small. Small enough to be live and in real hands in weeks, not quarters. Then we look at what people actually reach for, and build phase two from evidence rather than the wish list.

It costs me revenue in the short term. It has never once cost me a client.