Response, Resolution and Repair Time

Time-based measures are the most quoted and the most manipulable in maintenance. The manipulation is usually in the definition rather than the data.

The three clocks

Response time. From report to arrival on site. What most agreements specify.

Resolution time. From report to the equipment working. What actually matters.

Repair time. Time the technician spent working. Relevant to billing, not to operations.

A four-hour response with a three-day parts wait is a three-day outage, reported as excellent performance. This is the standard gap between what is measured and what the operator experiences.

Measure resolution

From report to verified working, confirmed by the site rather than by the vendor closing the ticket.

Broken down by priority, because a P1 and a P3 have different expectations.

With parts waits identified separately, so that a vendor is not penalised for a manufacturer's lead time and an operator can see where the delay actually sits.

Excluding time waiting for operator approval, which is your delay and should be measured separately — approval latency is frequently a larger contributor than anything the vendor controls.

The definitions that get gamed

Clock starts on dispatch rather than on report. Removes triage delay from the measure.

Clock stops on arrival. The response measure.

Clock pauses while awaiting parts, which can be legitimate and can also hide a vendor who never orders anything promptly.

Ticket closed and a new one opened for the return visit, which resets the clock and also flatters the repeat rate.

Agree the definitions in writing with the vendor, including what pauses the clock and what constitutes closure.

Averages hide the problem

A median is more informative than a mean for maintenance timing, because a few very long outages distort the average.

Report the distribution, or at minimum the median and the worst case.

Report the proportion within commitment rather than the average time. "Ninety-one percent of P1 within four hours" is more actionable than "average 3.8 hours", because the failures are what you want to see.

Look at the tail. The work orders that took three weeks are where the process broke, and they are invisible in an average.

Downtime is the operator's measure

Vendors measure their own performance. The operator should measure the equipment.

Hours the asset was unavailable, from failure to verified working.

Whether service was affected, as a flag.

Consequential loss, estimated.

This is the number that connects maintenance to the business, and it is the one that makes the case for redundancy, faster response commitments or replacement.

What good looks like

Realistic targets for a well-run multi-site operation with contracted maintenance:

P1 response within four hours, including out of hours, at high nineties percent compliance.

P1 resolution same day for the majority, with parts-related exceptions identified.

P2 resolution within two business days.

P3 batched and cleared within two weeks.

Repeat visits under one in ten work orders.

These are achievable and they require agreements that specify them and measurement that verifies them.

Where the delay usually is

Measured honestly, the delays in most operations distribute as:

Approval latency. The operator's own decision time, frequently the largest single component and never measured.

Parts sourcing.

Vendor capacity, particularly seasonal.

Diagnosis, where a repeat visit was needed.

Actual travel and repair, which is usually the smallest part.

The implication is uncomfortable and useful: the fastest available improvement in resolution time is usually raising the approval threshold so that routine work proceeds without a call, rather than pressing the vendor to drive faster.

Measuring your own approval delay

The component of resolution time that operators control entirely and almost never measure.

Record the time from work order raised to approval given, where approval was required.

Report it separately from vendor performance.

The finding is usually uncomfortable: approval latency frequently exceeds vendor response time, particularly for anything above the threshold, and particularly at weekends.

The fix is structural rather than exhortative. Raise the threshold so that routine work proceeds without a call. Delegate authority to a named person at each region rather than routing everything centrally. Define a standing authorisation for defined emergency conditions.

An operator who reduces approval latency by a day has improved resolution time more than any vendor negotiation would achieve, at no cost.