Lead Time to Learning: the number your operations review is missing
One number, for one pilot, with the waits written underneath it. How to measure the days between an accepted idea and the first live signal, and why you should never bonus it.
Adapted from Chapter 5 of The Seven Frictions
A CIO I worked with asked her platform product manager one question before a Thursday operating review: how long does it actually take us to learn something? He came back the next morning with one pilot and a whiteboard.
Support had wanted to put AI-generated reply suggestions in front of five percent of agents for a week, with full logging and an off switch. Leadership accepted the idea on a Monday at nine in the morning. The first suggestion reached a real agent six weeks later. Forty-two calendar days.
Four of those days were build and test. The other thirty-eight were waiting: for a slot on the shared staging cluster, for a masked sample of support tickets, for the weekly change board, for the monthly release window. Nobody in that chain had missed a deadline. Measured against their own clocks, every one of them had performed.
A question about the system
What changed in the room was the question. Until that morning the CIO had been asking why a particular team was slow, which is a question about people, and questions about people get answered with excuses or with a new manager. After it she asked what makes learning slow here, which is a question about the operating model. That one gets answered with a backlog.
One page per initiative
Lead Time to Learning is the measure: the calendar days between the moment an idea is accepted and the moment the first live signal arrives. The Latency Scorecard is how you carry it into a room where decisions get made, and the discipline is that it stays one page.
Beneath the total, split the elapsed time into four waits: approvals, environment, data and change controls. Score each from zero to five, where zero is good and five hurts. The four scores add up to a Latency Index between zero and twenty. End the page with two improvements, each with a named owner and a date.
Splitting one total into four waits changes who is in the argument. A single number invites the business to say IT is slow and IT to say the business keeps changing its mind. Four lanes put a name on each bar. The data steward owns one, the platform team another, the change manager a third.
Three rules that keep it honest
- Start with one pilot, not a dashboard. One number leadership can move beats twenty it can admire.
- Write the anchors down. Scores only matter if two teams in different buildings would land on the same number. If you are torn between a two and a three, pick one and move on.
- Never bonus the number. Lead Time to Learning is a property of the system, not a performance rating for a team. Review it monthly, demand a paragraph with it, and keep it out of compensation.
In most organizations the two largest bars turn out to be approvals and change controls, which are usually the same bar seen from opposite ends. The next move is a fast lane for changes that are small and reversible, and that is where the book goes next.