• Aug 5

The Wrong Test for Done

In strategic R&D work, a document gets called done when it looks done. A clean deck. A thorough-looking analysis. A senior name on the cover. A deadline met.

Every one of those is about how the document looks. None of them asks whether it works, whether someone could take the decision it was built for and actually make it from what is on the page. That question is the one that matters, and it usually goes unasked.

Watch what actually closes these reviews. A technology screening is accepted because the expert who ran it is trusted, and trusting is easier than checking. A deck is signed off because the deadline arrived and something had to be signed off. A review closes because the meeting closes. Each of these is a real signal. None of them tells you the work is ready to commit the portfolio to.

The idea of a definition of done came from software, and R&D borrowed it for good reasons. It does not transfer cleanly. In software the artifact is the deliverable: the code ships, the tests pass, the thing runs, and once it runs the work is done, because the running code is the point. A strategy document is different. The document is not the deliverable. The decision it informs is. Measure the document against its own polish and length and internal consistency, and you have measured the wrong thing carefully.

A document is finished when it fits the decision it serves. That is why a three-page note the team can act on can be more finished than a forty-slide deck that leaves the choice exactly where it started. The extra slides are more work, and none of that work moved the decision.

The proxies are not foolish. A trusted expert usually is right. A polished deck usually did take real effort. Nobody can re-derive every analysis before committing, and nobody should try. The proxies are useful as evidence. They go wrong only when they replace the one test they cannot stand in for: whether the decision can be made on the work. Let the expert's reputation set how closely you read the work. It cannot tell you the work is done.

Some decisions are too uncertain for the work to settle. No amount of analysis makes the choice safe, because the uncertainty has not resolved yet. A finished-looking deck does not change that. It only hides it. Sometimes the most finished thing you can produce is a clear account of what you still do not know, handed to the person who has to decide anyway.