Lead time: how to define it, calculate it and shorten it

Before anyone uses a lead time number in a dashboard, a quote or a supplier review, write these down:

• Start event: The action that starts the clock, such as order placed or request submitted.
• End event: The action that stops it, such as delivered, shipped or accepted.
• Unit and calendar: Hours, calendar days or workdays, and whose work calendar.
• Scope: Which steps and which waiting periods sit inside the measure.
• Status: Whether the figure is an estimate, a promise or an actual result.

Use the same five answers everywhere the number appears. Most arguments about lead time are really arguments about boundaries.

Lead time is a sum of different kinds of time, not one steady activity. Splitting it shows whether the delay comes from the work or from the gaps between steps:

Not every process has all five, and some have more, such as sourcing and supplier onboarding in purchasing. Include only what sits inside your chosen scope. Where stages overlap, for example packing that starts before the last parts are made, draw the flow first so you do not add the overlap twice.

Teams use different boundaries because they make different decisions. Naming the type keeps a planning estimate from being mistaken for a customer promise:

Treat these as sample definitions to adapt, not fixed rules. Check whether your supplier stops the clock at shipment, arrival or acceptance, and decide whether procurement starts when someone notices a need, raises a request or releases the order.

In purchasing, ARO stands for after receipt of order. A quote that says four weeks ARO starts the clock when the supplier receives the order. It does not say what happens when the four weeks are up.

Ask three questions before you plan around it:

• What ends the period? Ready to ship, shipped, delivered, or accepted after inspection.
• How are days counted? Calendar days, business days, or weeks with holidays excluded.
• What counts as receipt? Some suppliers start the clock only when full specifications or payment arrive.

Record the agreed meaning on the purchase order. An unclear quote should not become a firm customer promise.

For work already done, subtract the start timestamp from the end timestamp:

Lead time = end timestamp − start timestamp

An order received at 10 a.m. on April 3 and delivered at 10 a.m. on April 15 has a lead time of twelve calendar days. If your report uses workdays, recount with the agreed work calendar. Do not relabel the same twelve days.

For a forecast of a process whose stages do not overlap, add the stages. This illustrative order totals eleven days:

Only two of the eleven days are spent making the product. Once the work is done, the timestamp gap remains the direct measure; the stage sum is a plan.

Average lead time is the total lead time of finished orders divided by the number of finished orders. On its own it hides the orders customers complain about.

Take five illustrative orders that took 6, 7, 7, 9 and 16 days. The average is 9 days, but the median is 7. The gap points straight at the 16-day order, which is the one worth studying.

Decide in advance how to treat partial deliveries, cancellations and reopened requests, and keep the rule fixed. A change in counting rules can look like an improvement that never happened. The cycle time guide explains percentiles in more depth.

These clocks answer different questions, and teams do not always use the names the same way. Check the definition before comparing two reports:

A faster production step does not prove that customers wait less. Compare start-to-finish results before claiming an end-to-end gain. The cycle time guide has a side-by-side table with takt time, and the takt time calculator works out the pace.

Delivery dates are milestones, not durations. Arrives Friday names an event; five workdays names an interval. In customer messages, state both where you can, along with the event that starts the interval.

Build a small set around the decisions your team actually makes:

• Average lead time: A summary of finished orders in one defined group.
• Median lead time: The middle value when orders vary a lot.
• 85th or 90th percentile: The time most orders finish within, for promises.
• On-time rate: The share finished by the date promised.
• Stage durations: Time in approval, queue, production, transport and other steps.
• Open-order age: Elapsed time for unfinished orders, kept apart from finished ones.

Start with the end-to-end figure, one measure of reliability and enough stage detail to study exceptions. Add more only when someone can name the decision it will support.

Start with the real process, not the documented one. A value stream map shows where work moves, stops and waits:

1. 1

   Pick a comparable group. One product family, supplier route or service type. Mixing custom and standard orders blurs both.

2. 2

   Check the timestamps. Confirm the dates record real events, not data entered late in a batch.

3. 3

   Follow individual orders. Rebuild each path through approvals, queues, processing and completion.

4. 4

   Separate work from waiting. Mark the waits you want to study more closely.

5. 5

   Look at the unusual cases. Note whether long orders share missing information, missing material, rework or a custom spec.

6. 6

   Choose one change to test. Give it an owner, a review date and the measure that will show whether it helped.

Confirm causes with the people doing the work. A timestamp shows where an order waited, not why. A long approval gap may be a queue, missing information or a required review.

Shorter lead time comes from better flow, not from people rushing. Four places usually repay a look:

Keep quality checks visible inside the flow. A shorter headline number that comes with more defects or incomplete deliveries is not an improvement.

These illustrative cases show how the boundary changes the conclusion:

1. 1

   Quoting production time as delivery time. The customer also waits for approvals, queues, packing and transport.

2. 2

   Mixing boundaries. Comparing a supplier's order-to-dispatch figure with your order-to-receipt figure.

3. 3

   Reporting only the average. A few very late orders disappear into it.

4. 4

   Ignoring unfinished work. Finished-order statistics look better as the oldest orders get stuck.

5. 5

   One target for every order. Routine and custom work may need separate targets, with the reason written down.

6. 6

   Treating a target as history. A goal is what you want; history is what happened. Do not swap one for the other in planning.

Write a one-page definition with the start event, end event, calendar, included steps and exception rules, and ask sales, purchasing, operations and service to use it. Then hold a short regular review:

Pick the goal before ranking projects. A lower average may matter for one customer group, while fewer very late orders matters more for another. Lead time works with Little's Law: with the same throughput, less work in the system means less time in it.