Decision velocity is a balance-sheet item

When a decision waits, everything behind it waits too: the capital it would release, the hire it would approve, the price change it would set. That waiting has a price, exactly as warehoused stock does. Across 40 organisations we traced the last sixty significant decisions each had made — from the moment the decision became necessary to the moment it was communicated — and priced the delay against the value case of the work each decision was holding up.

The result is a distribution nobody enjoys seeing their position on. The median organisation held nine working weeks a year of decision queue, priced at $6.8M in delayed value and rework. The slowest quartile held more than twice that. The variable that best predicted speed was not industry, size or management fashion. It was whether the decision had a single named owner. We should say what the pricing rests on. It uses each decision’s own value case, so an organisation in the habit of writing optimistic value cases will look like it is losing more than it is.

Slowness is not evenly distributed. Single-owner decisions in the sample moved in a median 4 days; cross-functional decisions took 34. The difference is not analytical difficulty, and many of the slow decisions were analytically trivial. A decision without one owner has to build consensus as it travels, and consensus-building is the most expensive transport a decision can take, because every stop on the route adds a meeting, a version and a wait.

Strikingly, 71% of measured latency was attributable to missing authority rather than missing information. The organisations were not waiting to know more. They were waiting for someone to be allowed to decide. That is the finding clients contest hardest, usually by arguing that the extra analysis was genuinely needed, and the timestamps in their own approval trails rarely support them.

Treating latency as a cost line changes behaviour in a way that exhortations to be more decisive never do. Once the queue is priced, it competes for management attention like any other cost. The repairs that follow are unglamorous and effective: one named owner per recurring decision class, a fixed forum with the authority to decide, and a quarterly re-measurement to stop the queue growing back.

The measurement takes under a week and is published as a public instrument. What it costs a client is access to the minutes, the approval trails and the people who were waiting. The number, once seen, cannot be unseen, and that is the point of publishing the method rather than selling the finding.