Every MSO tracks average cycle time. Most are measuring it wrong.
The problem isn't the math. It's that averages hide the outliers—and the outliers are where the problems live.
A shop averaging 9 days sounds healthy. But if that average includes a 45-day repair, you have a problem that isn't showing up in your metrics.
The Average Lie
Consider two shops with identical averages:
Shop A - 10 day average:
20 repairs: 8, 9, 9, 9, 10, 10, 10, 10, 10, 10, 10, 10, 10, 11, 11, 11, 11, 12, 12, 12
Shop B - 10 day average:
20 repairs: 5, 5, 5, 6, 6, 6, 7, 7, 7, 8, 8, 8, 9, 9, 10, 10, 12, 15, 25, 42
Same average. Completely different performance.
Shop A is consistent—every repair within a tight range. Shop B has a few repairs dragging the average up. One 42-day repair is masking what would otherwise be a 7-day average.
Better Metrics Than Average
Median (P50)
The median is the middle value. Half your repairs are faster, half are slower. Unlike averages, medians aren't skewed by outliers.
- Shop A median: 10 days
- Shop B median: 8 days
P90 (90th Percentile)
90% of repairs are completed within this time. This is your "worst case" metric.
- Shop A P90: 12 days
- Shop B P90: 25 days
Now Shop B's problem is visible. Most repairs are fast, but 10% are taking over 25 days.
Distribution Shape
Plot your cycle times as a histogram. A healthy distribution has a tight cluster around the median with a short tail. An unhealthy distribution has a wide spread with a long tail stretching right.
Finding the Outliers
Outliers aren't random bad luck. They're symptoms of systemic problems:
Parts Delays
The most common cause. Filter for repairs with cycle time > 20 days and check how many are marked "waiting on parts."
Supplement Loops
Repairs with 3+ supplements are almost always outliers. Compare cycle time to supplement count.
Forgotten Vehicles
Repairs that fall through the cracks. Look for repairs with long gaps between last activity and completion.
Severity Mismatches
A $15k repair taking 20 days is different from a $2k repair taking 20 days. Segment by repair severity.
The Right Way to Track Cycle Time
Level 1: Network Summary
| Metric | This Week | Last Week | Trend |
|---|---|---|---|
| Median cycle time | 8.2 days | 8.5 days | ↓ |
| Average cycle time | 10.1 days | 10.4 days | ↓ |
| P90 cycle time | 18 days | 19 days | ↓ |
| Outliers (>25 days) | 12 | 15 | ↓ |
Level 2: Shop Comparison
Rank shops by median cycle time, but flag those with high P90s. A shop with good median but concerning P90 has outlier problems worth investigating.
Level 3: Outlier Drill-Down
For flagged shops, list the specific repairs causing problems with reasons: parts delay, adjuster delays, customer unresponsive.
The 37-Day Shop
Real example from a 100+ shop MSO:
During a routine dashboard review, leadership noticed one shop with an unusually high average: 15 days when the network was running 9.
Drilling down, they found:
- Median was 8 days (reasonable)
- P90 was 37 days (terrible)
- 6 repairs over 30 days in the past month
- All 6 were waiting on the same parts supplier
The fix was simple: switch suppliers. 30+ day count dropped from 8 to 1 within a month.
They caught it in week 3, not month 3.