metricDelivery & agility

Lead Time

Lead time measures elapsed time from a request or commitment point until delivery, capturing both queue time and time spent actively completing the work.

Download Markdown (.md)

01What it means

Lead time is the total elapsed time from a defined upstream event—often a customer request, backlog commitment, or order—to a defined delivery event. Unlike cycle time, its boundary commonly includes time before active work begins. The exact meaning is contextual, so reports should name both endpoints and the calendar used.[1][2]

Total elapsed delay from an upstream request to delivery Lead time contains queue time before work plus the measured delivery cycle
Lead time = delivery − request or commitment
Upstream start
Customer request, order, or commitment point
Queue before work
Waiting until active delivery begins
Cycle time
Active work and all internal waits to finish
Delivery boundary
Observable release or fulfilled outcome

Always name both endpoints and the calendar. Analyze distributions and higher percentiles because long waits are part of the customer experience.

02Why it matters

Lead time reflects the wait experienced by a requester more closely than an internal activity measure. It helps teams understand responsiveness, expose queues before work starts, set evidence-based expectations, and see whether prioritization or intake policies create delay even when execution itself is fast.

03When to use it

Use lead time when customers or stakeholders care about request-to-delivery duration, when a substantial queue exists before work starts, or when comparing service responsiveness over time. Choose the starting event that represents a real promise or demand signal rather than whichever timestamp is easiest to extract.

04How it works

  1. Select a meaningful request or commitment event and an observable delivery event for one class of work.
  2. Capture the two timestamps consistently, including waiting and active time within the chosen boundary.
  3. Analyze the distribution with medians and higher percentiles, preserving long observations rather than trimming them away.
  4. Separate cohorts when urgency, item type, workflow, or service policy makes their lead-time behavior different.
  5. Use queue and cycle-time evidence to locate where delay accumulates, then test policy or workflow changes.

05A practical example

Example

A research-operations group measures from an approved study request to delivery of a findings readout. The median lead time is 18 days, although hands-on research takes six. Most delay occurs while requests await participant-recruiting capacity. The group introduces intake limits and clearer priority rules, then monitors the full distribution.

06Where it breaks

Lead time does not explain why an item took as long as it did, and it may mix queueing, rework, active effort, and external delay. Comparisons become misleading when endpoints or service classes change. Shortening lead time can also be harmful if it encourages premature commitment, reduced discovery, or lower quality.

Common mistakes

  • Calling a measure lead time without documenting the request and delivery events that define it.
  • Excluding queue time even though responsiveness to the requester is the question being asked.
  • Comparing unlike work classes or using one promised date for every percentile of the distribution.

07Sources

  1. Cumulative Flow, Lead Time, and Cycle Time GuidanceMicrosoft Learn · Technical guidanceCumulative Flow, Lead Time, and Cycle Time Guidance
  2. About Kanban BoardsMicrosoft Learn · Technical documentationAbout Kanban Boards
View saved topics

Start typing to search titles, definitions, categories, and frameworks.