Shipping the boring version first

I spent six weeks on a scheduling interface that let you drag a range across a week, snap it to your working hours, and see conflicts resolve as you moved. It was the best thing I built that year. About eleven people used it.

The thing they actually wanted was a list of times with checkboxes next to them, which I built afterwards, in an afternoon, in a state of some annoyance.

What the boring version buys you

Not speed, though it is faster. What it buys is the only information that matters: whether anyone wants the feature at all, and what they try to do with it when they have it.

Those two answers change the design more than any amount of thinking does, and you cannot get them from a prototype, because a prototype only tells you whether people understand the thing. It never tells you whether they will come back to it on a Tuesday.

The part nobody says out loud

Shipping the boring version is uncomfortable in a specific and slightly humiliating way. It looks like you did not try. Colleagues who have not seen the plan assume that the list of checkboxes is the design, and you have to sit with that for a few weeks.

The way through is to be explicit: this is version one, here is what I am watching, here is the date I will decide. That reframes the boring version as an instrument rather than an opinion, which is what it actually is.

Where it does not apply

It does not apply when the interaction is the product. If you are building a drawing tool, the boring version of the canvas is not a smaller version of your idea, it is a different and worse product, and you will learn nothing from it except that people do not like bad drawing tools.

It also does not apply to anything that is hard to walk back: pricing, data models, URLs, anything a customer will build a habit on. Ship those slowly and carefully. Everything else, ship badly and early, and let the second version be the one you are proud of.

The scheduling thing

I did eventually get to build the drag interaction. It went out fourteen months later, on top of the checkboxes, for the people who had by then done the thing enough times to want it faster. About a third of them use it. That is a much better number than eleven, and it is the same six weeks of work — just spent after the question was answered rather than before.