Main idea

Six months after a PRD, the organization that has to decide what comes next digs up the document and goes looking for the rest: it re-interviews the same people, asks again for explanations, rereads notes, reconstructs the arguments. What gets counted then is the days spent — and that is the cheapest part of the operation.

A reconstruction doesn't give back the original. It keeps the arguments that tell well in a meeting, loses the faint signals raised by support, forgets the sales objections nobody has replayed since, and drops the alternatives that were ruled out along with the reason for ruling them out. So the next decision is made on poorer material than was available for the first.

The cost of scattered knowledge is therefore of two kinds: a delay, visible and quantifiable, and a degradation in the quality of the decision, invisible because you never see the decision you would have made with the complete context.

Why it matters

This changes the argument you defend capitalization work with. Presented as a time saving, it is compared with other time savings and almost always loses the arbitration; presented as a condition for the quality of product decisions, it changes category.

It also explains why an organization that starts over may never notice. The delay is felt; the lost nuance isn't — an impoverished decision looks exactly like a good one.

Nuances and limits

A reconstruction isn't always an impoverishment: with six months' hindsight, new facts and additional customers, what comes back may be better than what was understood at the time.

And not every decision deserves this care: on a subject that won't come round again, deciding with an impoverished version of the context remains rational.

Open questions

• How do you measure the gap between the decision made and the one a complete context would have allowed, given that the complete context no longer exists?