Explore average time in a stable system. This calculation explains a relationship; it does not predict a task or guarantee faster delivery.
ILLUSTRATIVE AVERAGE
12 ÷ 8 × 5 = 7.5 working days
What the calculation means
Average time = average WIP ÷ throughput. The working-day display multiplies the result in weeks by the selected working days. Use consistent item definitions, system boundaries and observation windows. The relationship is commonly called Little’s Law.
Check the assumptions first
Use averages from a reasonably stable system; a growing queue is a warning.
Count the same class of work at entry, while active and at completion.
Keep blocked work inside WIP if it remains within the chosen boundary.
Do not mix story points with a count of delivered items.
This is an average, not a percentile, a service-level expectation or an individual promise.
A change is a hypothesis
Changing WIP while holding throughput fixed illustrates the model. A real change may also affect throughput, demand or quality. Observe the result rather than presenting the calculator as evidence of an improvement.
The WIP model above is an average relationship. To examine variation, enter durations from completed items using the same start and finish definitions. Use one number per line with a decimal point; no dates, commas or units. All entries must use one day convention. The seven example values below are fictional.
This summary needs JavaScript. The formula and worked example remain readable on this page.
Method and a checkable example
Sort the values. For proportion p, the zero-based position is p × (n − 1); interpolate linearly between adjacent values. This is the R7 convention described among the alternatives in NIST’s percentile reference. It differs from some other percentile conventions. For the example: n = 7, sum = 55, mean ≈ 7.857 days, median = 4 and sample p85 = 13.8 days. The display rounds to three decimals; calculations use unrounded values.
A sample percentile describes these completed observations. It is not an 85% guarantee for the next task. A few observations, a changed process or mixed work classes weaken any inference. Unfinished and cancelled work is not represented by this list; examine it separately to avoid hiding aging work. Do not mix working and calendar days. No input is sent or saved by this tool.