How do I track callbacks and warranty recalls properly?
Define a callback as a return visit to the same location and equipment within a set window, related to prior work. Do not rely only on someone remembering to tag it, because the tag is applied by the people it reflects on. Derive candidates from the data — same location, short interval, overlapping equipment or task codes — and have a human confirm. Then normalize the rate by job type before comparing anyone.
The definition has to come first
Callback means different things in different shops, and the disagreement is usually about the window and about causation. A return trip eleven days later for the same failure is clearly a callback. A return trip four months later for a different component on the same unit probably is not.
Pick a window from your own repair data rather than from a benchmark. Look at the distribution of intervals between consecutive visits to the same equipment and find where the curve flattens. That inflection is your natural boundary, and it will differ by trade.
Why tagging alone under-reports
The person best positioned to mark a job as a callback is often the person whose work is being called back, or a dispatcher trying to get the truck moving. Under time pressure and mild self-interest, the tag gets missed. Not maliciously — it is simply the least urgent thing on the screen.
Deriving candidates from structural evidence removes that pressure. Same location, within the window, overlapping equipment or task codes, and the second job carrying little or no revenue is a strong signal. Surface the candidates for confirmation rather than counting them automatically, because the false positives are real — unrelated failures do cluster in old systems.
The cost is bigger than the unbilled hour
- Lost capacity. The slot the callback consumed was a slot a paying job could have used. On a full board that is the largest component of the cost.
- Displaced margin. The original job keeps its healthy costed margin while the return trip's cost lands somewhere else, so both jobs look better than the pair actually was.
- Reputation. Callbacks correlate with the reviews you least want, and reviews feed the local search visibility you pay for elsewhere.
- Parts and warranty handling. Restocking, supplier claims and shop time that rarely make it onto any job record.
Normalize before you rank
Callback rate varies by job type, equipment age and system complexity far more than by technician skill. A technician working mostly on aging equipment in older housing stock will show a higher raw rate than one doing new installs, and ranking them together is simply wrong.
Compare within job type, or model the expected rate given the mix and look at the residual. Then treat the outliers as a starting point for a conversation, supported by what actually happened on those visits. That is the same discipline that applies to every other performance number in ServiceTitan reporting, and it is why evaluation data is framed to support human coaching rather than to hand down a verdict.
Topics: callbacks · warranty · quality · job data
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.