On this page
Define the return before measuring it
A return visit can concern workmanship, a failed component, an unrelated new problem, a promised follow-up or a customer preference. Those events have different causes and economic treatment. Record the actual work performed and the evidence for the classification. A second appointment alone does not establish defective work, and a manufacturer’s parts warranty does not automatically reimburse the contractor’s labor or travel.
Choose the population of original jobs completed in a stated period. Reconcile anonymous job references and completion dates to work orders and the ledger or dispatch control. A list containing only returns cannot establish the denominator. The existing garage-door example lists twelve return events across eight affected jobs in a separately stated 100-job cohort. It explains counting; the expanded example adds explicitly hypothetical dates and costs for a reproducible cost review.
Match the time available for returns to appear
Choose a follow-up horizon appropriate to the question and keep it explicit. Ninety days in the example is an assumption, not an industry standard or a warranty term. A job is mature for that analysis when its completion date plus ninety calendar days falls on or before the observation cutoff. A recent job with only fourteen days of observation is not comparable to a job followed for the full ninety days.
The callback tool separates mature jobs, incomplete follow-up and events outside the chosen horizon. Its matched-horizon rates and costs use only eligible mature jobs and their included returns. Later events stay visible in a separate bucket; they do not disappear from the evidence file. If no job is mature, the rate is unavailable rather than zero. A clean early snapshot cannot prove that the work will remain free of later problems.
Keep affected jobs, events and repeated returns separate
| Hypothetical mature cohort | Calculation | Result |
|---|---|---|
| Affected-job rate | 8 distinct affected jobs ÷ 100 completed jobs | 8% |
| Return-event frequency | 12 included events ÷ 100 jobs × 100 | 12 events per 100 jobs |
| Additional events beyond each first return | 12 events − 8 affected jobs | 4 additional returns |
| Distribution | 92 jobs with none; 5 with one; 2 with two; 1 with three | 100 jobs / 12 events |
Do not call the twelve-event figure a 12% customer failure rate. One job can generate several visits. Preserve event references, dates, root-cause review and the original job link so a reader can reproduce both counts. Compare job types separately where installation, repair and routine maintenance have different scopes. A blended change can reflect service mix, customer reporting or coding practices rather than a change in workmanship.
Reconcile cost and recovery to their records
ASQ’s cost-of-quality framework separates prevention and appraisal from internal and external failures. It is useful for organizing questions, but it provides no callback percentage or contractor reserve for this business. Build the company’s review from time records, payroll allocation, issued parts, vehicle costs, refunds, supplier credits and the actual promises to customers.
| Original cost example | Amount | Evidence convention |
|---|---|---|
| 18 paid person-hours × assumed $60 loaded rate | $1,080 | Allocation method and denominator stated. |
| Parts issued to returns | $600 | Exclude parts already counted elsewhere. |
| Other direct travel/nonlabor cost | $240 | Do not include vehicle allocation twice. |
| Gross observed cost | $1,920 | Subtotal before recovery. |
| Credits actually received | −$300 | Matched to the same return population. |
| Net observed cost | $1,620 | $16.20 per mature completed job. |
The example has $202.50 net cost per affected job and $135 per event. Those denominators answer different questions. The loaded labor figure is not automatically cash that could have been saved: payroll may remain fixed if a return is avoided. Pending manufacturer reimbursement belongs in an evidence queue, not in the received-credit line. A credit above gross cost remains a negative net result requiring a timing or population review.
Test coding and sample coverage
Reconcile the job manifest to the complete-population control and disclose differences. Check duplicate job IDs, repeated event exports, missing links, returns before completion and events after the observation cutoff. Trace a deliberate sample to original work orders, parts issues and payment or credit records. Include repeated-return jobs and unusual credits rather than choosing only clean cases.
Use anonymous references in working files. Keep customer identities, employee-level records and sensitive complaint correspondence in the controlled data room. A root-cause label is an assertion requiring review. This descriptive tool does not rank technicians, prove causation, predict future claims or determine a legal warranty obligation.
Carry the result into pricing and closing once
Observed historical cost, a future open obligation, a pricing allowance and an accounting reserve are separate schedules. An unfinished promised repair needs a supported completion plan; a historical event already expensed should not be deducted again as if it were unpaid. Identify whether any parts, credits, refunds or remaining work are included in the purchase agreement, working capital or another adjustment.
Use the open customer-obligations guide for actual promises and the pricing model for a separately supported allowance. Keep one issue reference across the earnings bridge, reserve discussion, agreement schedules and cash plan. An observed $1,620 cost is not automatically a price deduction, a future reserve or $1,620 of avoidable cash savings.
Companion resources
Use the printable or editable materials alongside the guide.
Mature-job and return-event CSV exampleCallback cost and recovery evidence registerRelated decisions
Sources and scope
Source links support the specific topics noted. Examples and analytical frameworks are illustrative; confirm current requirements and transaction facts with the responsible advisers.
- ASQ — Cost of Quality
Prevention, appraisal and internal/external failure categories. Our job-cohort model and costs are original examples, not ASQ contractor benchmarks, predicted savings or warranty-reserve guidance.
Continue your research
Browse the research library · Calculators and downloads · Trade profiles