An episode model: everything from the first missed call to the first real recovery for one counterparty is ONE episode, regardless of how many calls happened in between. Episodes carry the miss cause, who was accountable, whether somebody else covered, and whether the close was observed or self-attested.
Also called: speed to lead · response time · callback time · how long before we ring back
One number: how long it takes you to get back to somebody you missed
The clock starts at the first missed call from a person and stops at the first recovery this system can actually see — a call placed back to them from inside the product, or a text sent to them. Everything in between is one episode, so four calls from one number before one callback reads as one problem rather than four failures and a success. Speed to lead is the number that decides who gets the job, and most builders have never seen theirs.
Texting a missed caller counts as calling them back
For this trade a text is often the better move, and it closes the debt exactly the way a call does, timed to the moment it went out. Without that, the rep who did the right thing looks like the one who ignored the lead.
The number cannot be improved by dialling more or by tapping a button
When somebody settles a call from their own cell they tap ‘I got this one’ — which clears the queue but is excluded from the response times, so a manual close can never fake a fast number. Dial counts measure nothing about whether a customer got called back, and are not scored here.
A miss nobody could have answered is not counted against anybody
Misses with no owner — a line with no forward configured, a call after hours — are marked as such and never enter the figures. A rep who covers for a colleague is recorded as covering, not as an exception to explain. A metric that makes a good employee look bad gets the whole dashboard distrusted, permanently, which is why the rules are written this way.
- 1A unique partial index on (org, kind, counterparty) WHERE status='open' makes one-episode-per-person structural, not a read-time grouping trick
- 2Keyed on the COUNTERPARTY alone, so a rep who misses on his own line and rings back from the shop's main line still counts as recovered
- 3Closed by an outbound connected call, an outbound text, or a manual tap
- 4Miss causes that are nobody's fault (no forward configured, after hours) are marked unqualified and never scored
- 5accountable_user_id may be NULL — an unowned main line's miss is owed by the account
- 6self_attested closes are EXCLUDED from the latency distribution
Speed of callback is the number that decides whether a lead becomes a job, but measuring it naively makes it useless. Four missed calls from one person before one callback is one problem, not four failures and a success — a per-call count would report it that way, which is both wrong and demoralising to the rep it lands on. The measure runs per person, from the first miss to the first genuine recovery, whether that is a connected callback or a text back. A callback made from a personal cell is invisible, so a manual tap-to-close is excluded from the timing rather than allowed to fake a fast number, and where the owner of a miss is uncertain it is attributed to nobody rather than to the wrong person.
- No measurement of speed-to-callback
- Metrics that punish a covered save or an unavoidable miss
- Numbers improvable by tapping a button
See it on your own jobs
Twenty minutes, your numbers, no slide deck. We’ll build one of your real buildings in front of you and send you the estimate link at the end — yours to keep either way.
or keep browsing features →