What should the denominator be when I calculate estimate close rate?
Estimates presented in a cohort period, not estimates open today and not estimates created today. Close rate equals estimates sold divided by estimates presented from the same cohort, measured after enough time has passed for that job type to resolve. Using a live pipeline as the denominator makes recent months look artificially bad, because those estimates have not had time to close yet.
The metric most companies compute wrong
The common version is sold this month divided by presented this month. It mixes cohorts: some of the sales came from estimates presented weeks ago, and some of this month's estimates will sell later. In a growing company this understates close rate. In a shrinking one it overstates it. Either way it moves for reasons that have nothing to do with selling.
The fix is cohort accounting. Tag every estimate with its presentation date, then measure what fraction of that cohort had sold by day thirty, day sixty, and so on.
Why cohorts change the conversation
A cohort view separates two things that get blamed on each other: how many estimates eventually close, and how fast they close. A team can have a healthy final close rate and terrible speed, which shows up as cash flow problems and technicians re-driving to the same houses. Another team can close fast on the easy ones and never revisit the rest.
Plot cohorts side by side and you see drift immediately. If the day-thirty number is falling while the day-ninety number holds, follow-up got slower but selling did not get worse.
Exclusions worth making explicit
- Estimates created for internal or accounting reasons. Placeholder and re-quote records inflate the denominator without ever being presented to a customer.
- Multiple option estimates for one decision. Good, better, best presented as three records is one decision, not three. Count the decision or your close rate is roughly divided by three.
- Warranty and no-charge work. These resolve on a different logic and belong in their own bucket.
- Estimates on jobs that were never dispatched. If nobody ever stood in front of the customer, that is a scheduling failure, not a selling failure.
Tie it to something that moves
Close rate on its own is a scoreboard. It becomes useful when broken out by the dimensions you can act on: job type, technician, lead source, and whether the estimate was presented on the first visit or a return trip. First-visit presentation almost always closes better, which makes dispatch decisions a selling lever as much as a capacity one.
Joining estimate outcomes back to the campaign that produced the original call is where this stops being a sales metric and becomes a media budgeting input. That join is the core of what we build in revenue intelligence, on top of a normalized feed from the field service system.
Topics: estimates · close rate · cohorts · metrics
Have a version of this question about your own business?
The useful answer usually depends on which systems you run and how they're connected. That's a conversation, not a blog post.