In 30 seconds
- Measured from request to delivery, including all waiting.
- What customers experience as speed.
- Shorter, more predictable lead times are the goal.
Try it
Little's Law: average lead time = average WIP ÷ average throughput. It holds for averages over a stable period, so use it for direction, not precise promises.
Lead time vs cycle time
| Lead time | Cycle time | |
|---|---|---|
| Starts | When requested | When work starts |
| Ends | When delivered | When work finishes |
| Who cares | Customers | The team |
Definitions vary between teams; agree yours and use it consistently. DORA's "lead time for changes" runs from commit to production.
Measuring and forecasting
Record start and end dates for each item and plot them on a scatter chart. Percentiles are useful: "85% of items finish within 9 days" is an honest forecast for a new request.
Reducing it
- Limit work in progress
- Split work smaller
- Remove waits and handoffs; see flow efficiency
Myth vs reality
The AI angle
AI shortens coding time, but lead time often stays the same because work waits in review, testing or approval. Measure end-to-end lead time to see where AI actually helps.
FAQ
How is lead time linked to WIP?
Little's Law: lead time = WIP / throughput, on average.
What's a good lead time?
Whatever meets customer needs; aim for shorter and more stable.
Should we track it per item type?
Yes; bugs and features often differ.