Repair cycle time is calendar days between the moment a repair need is raised and the moment the vehicle is back in service. Not shop days. Not business days. Not the labor hours on the invoice.

Cycle time = (date and time back in service) minus (date and time the repair request was created).

That definition is deliberately unforgiving, because every softer version of it hides the part of the process you control. A shop that books a vehicle in on Wednesday and finishes it Friday will tell you the repair took two days. If the request came in the previous Thursday, it took eight.

Measure it properly or do not bother

Three decisions determine whether your cycle time number is useful or decorative.

Start the clock at the request, not at the shop

The trigger is whichever came first: the driver reported it, the telematics fault fired, the damage photos landed, or the manager opened the RO (repair order). Anything later and you have measured the shop instead of the process.

Stop the clock at back in service, not at repair complete

A vehicle finished Thursday afternoon and collected Monday morning cost you four days. Those days are real, they are usually yours, and dropping them flatters the number by exactly the amount you could most easily fix.

Report median and 90th percentile, not the mean

One waiting-on-a-frame-rail repair drags an average until nobody believes the report. The median tells you what normal looks like; the 90th percentile tells you how bad the tail gets, and the tail is where your capacity actually goes.

The four segments

A single cycle time number tells you there is a problem. It does not tell you whose. Split every repair into four segments with explicit timestamps, and the argument ends.

Intake to estimate

Starts when the request is created. Stops when a priced estimate is in hand. This segment covers triage, shop selection, transport or tow, and, on collision work, the photo or desk estimate. It is where fleets lose days silently, because nothing appears to be wrong: the vehicle is simply not anywhere yet.

What good looks like: hours for a drivable mechanical job, one business day when transport is involved. If this segment regularly runs past two days, the problem is usually that nobody owns shop selection and the request sits in an inbox.

Estimate to approval

Starts when the priced estimate arrives. Stops when written authorization reaches the shop. This one belongs entirely to you, costs nothing to fix, and is frequently the largest segment in the whole cycle.

What good looks like: near zero for work inside policy and under your authorization threshold, and under four business hours for anything that needs a human. Track supplements separately, because a supplement restarts this loop and a second approval delay hurts more than the first (the vehicle is already torn down and occupying a bay).

Approval to shop-in

Starts at authorization. Stops when the vehicle is physically in a bay with work commenced. This is parts lead time, shop backlog, transport, and driver availability colliding.

What good looks like: one to two days when parts are on hand. If this segment is your largest, routing is the problem, not the shop. You sent the vehicle somewhere that had a good rate and no capacity.

Shop-in to complete

Starts at work commenced. Stops when the vehicle is repaired, quality checked, and ready. Severity drives the absolute number, so do not benchmark shops against each other on raw days. Benchmark them on promise accuracy instead: the gap between the ECD (estimated completion date) they gave you at shop-in and the day they actually delivered. A shop that says eight days and delivers eight is worth more than one that says four and delivers seven.

Where the days actually are

Take a 40-van fleet running six repair events a month at an 11-day average. Split it out and it typically does not land where people assume:

• Intake to estimate: 2.5 days
• Estimate to approval: 3 days
• Approval to shop-in: 2 days
• Shop-in to complete: 3.5 days

The shop owns 3.5 of 11 days. Everyone argues about the shop.

Now cut approval from three days to half a day, which requires no new vendor, no capital, and no technician. That is 2.5 days per repair, 72 repairs a year, 180 vehicle-days returned. On a 40-van fleet that is roughly half a van of capacity recovered for the cost of rewriting an authorization rule.

Use your own numbers rather than these. The point is the shape: the segments you control are usually bigger than the segment you complain about.

How to cut each segment

• Give intake a single front door. One channel, whether that is a driver text, a QR scan, a telematics trigger, or an FMS (fleet management system) push, so requests stop landing in four inboxes with three different formats.
• Route on capacity and capability, not on rate alone. A shop with the right tooling and an open bay beats a cheaper shop with a nine-day queue on almost every repair you will ever write.
• Raise your no-questions authorization threshold and audit above it rather than approving below it. Most fleets have their threshold set where it was five years ago and pay for the difference in days.
• Pre-authorize predictable categories outright. Brakes, tires, batteries, and PM (preventive maintenance) work rarely need a human reading a line item.
• Push for teardown before estimate on collision work where the damage is likely to hide more. Paying for teardown early is cheaper than discovering a supplement on day six.
• Put a named owner on the return leg. Vehicles finished on Friday and collected on Monday are a scheduling failure, not a repair failure.

What to do Monday morning

1. Pick the last 50 completed repairs and reconstruct four timestamps for each: request created, estimate received, approval sent, back in service. If you cannot reconstruct them, that is finding number one.
2. Calculate the median and the 90th percentile for the total and for each segment.
3. Find your largest segment. Fix that one. Ignore the others this quarter.
4. Publish the segment breakdown to your shops. The ones who have been carrying blame for your approval delays will become considerably more cooperative.
5. Set a 90-day target on the single segment you chose, and re-run the same 50-repair analysis at the end of it.

Frequently asked questions

What is a good repair cycle time for a fleet?

There is no single right answer, because severity mix, geography, and parts availability differ too much between fleets. The useful target is your own median from last quarter minus a specific number of days, with a segment named as the source of the reduction. A fleet chasing a borrowed industry figure usually ends up measuring something different from what that figure measured.

How is cycle time different from downtime?

Downtime is any period a vehicle is unavailable, including waiting for a driver, sitting out of season, or held for compliance. Cycle time measures one specific repair event from request to return. Cycle time is the piece a repair process can actually be held accountable for.

Should weekends count in repair cycle time?

Yes. Calendar days are what your operation feels. Measuring business days lets a Friday afternoon handoff look free when it costs you a real Monday. If you want to understand shop behavior separately, track shop-in to complete in business days as a secondary view, not as the headline number.

Who should own cycle time inside a fleet?

One person, with the segment breakdown in hand. Cycle time spread across maintenance, operations, and procurement becomes nobody's number, and the segments each of those functions owns never get compared against each other.

Where ServiceUp fits

ServiceUp is the agentic repair platform for modern fleets, and the segments above are what it operates on. Agents handle intake, route the vehicle to a shop with the capability and the capacity, audit the estimate against your pricing, warranty, and maintenance policies, approve valid work in seconds, and keep the follow-up moving without a person chasing it. Across the fleets running on the platform, that has meant 32% faster cycle times. See how it works for fleets.