Shipping on a slow week

Four years working for myself, and the thing I am still worst at is reading a week where nothing shipped.

Two kinds of nothing

There is the week where the problem is genuinely hard and the apparent lack of output is the work — reading, discarding approaches, sitting with a data model that does not fit. That week looks identical from the outside to the week where I have quietly lost the thread and am avoiding a decision by refactoring something that was fine.

The tell, for me, is whether I can state the question I am currently answering in one sentence. If I can, the slow week is real work. If the sentence keeps sliding around, I am avoiding something.

Write the sentence down

So I write it down on Monday, on paper, and read it on Thursday. That is the entire system. It has caught more drifting weeks than any tool I have tried, mostly because a sentence written three days ago is unsentimental about what you have actually been doing.

The client version

The harder half is external. A slow week still needs an update, and "still thinking" is both true and unbearable to receive. What works is reporting the decisions rather than the output: what I ruled out this week and why. It is honest, it is genuinely informative, and it converts an invisible week into something a client can push back on.

Twice that has produced a reply along the lines of we do not care about that constraint at all — which would have cost me another week to discover on my own.

Not every week is recoverable

Some weeks are just bad, and the pretence that every one of them is secretly productive is its own kind of dishonesty. The useful response is to end them early rather than extend them, which took me an embarrassingly long time to learn and which I still get wrong most months.