Flow Efficiency: Where the 5–15% Figure Comes From — and Why No Normal Value Holds Up

Anyone working on lead times runs into the same figure sooner or later: flow efficiency in knowledge work is typically 5 to 15 per cent. It appears in conference talks, in training material, and in the documentation of the tools that measure it.

The reason to chase it down was practical. A calculator needed a reference value, and that value needed a properly stated origin. The further back the trail was followed, the vaguer it became.

What flow efficiency measures

Flow efficiency puts the time actually spent working on something against the total lead time:

Flow efficiency = work time ÷ lead time × 100

A work item that takes twenty working days from trigger to result, with two days of genuine work inside it, comes to ten per cent. The other eighteen days are waiting: for a sign-off, for an input, for an environment, for someone who is currently busy with something else.

That is what makes the metric interesting. It moves attention away from how fast individuals work and towards the hand-offs between them. Faster teams rarely shorten the work itself — they shorten the waiting in between.

Looking for the primary source

The figure is credited to several different sources. None of the attributions survives scrutiny.

The one named most often is This Is Lean by Niklas Modig and Pär Åhlström (2012). The book treats flow efficiency as a concept — the value ranges are not in it.

The second trail leads to David J. Anderson, speaking at conferences. It rests on anecdote, and it carries a further problem: Anderson's own school today puts the range at 1 to 25 per cent, with best cases up to 40 — not 5 to 15. The same rule of thumb is credited in parallel to Daniel Vacanti. And the most widely quoted article on the subject, by Julia Wester in 2016, names no originator at all.

So anyone saying “it comes from X, not from Y” is swapping one weak attribution for the next. The question of authorship leads nowhere — and it is not the point. The point is that no published measurement exists for the 5 to 15 per cent.

What was actually measured

One published measurement does exist. Nick Brown analysed 63 teams at ASOS.com over twelve months and published the results in the ASOS Tech Blog in July 2023:

• 9 to 68 per cent measured flow efficiency across the teams
• 35 per cent on average
• below 20 per cent: low
• 20 to 40 per cent: medium
• 40 to 60 per cent: high
• above 60 per cent: very high

The gap to the rule of thumb is considerable. A team at 30 per cent sits well above the norm by the popular rule — and squarely in the middle band by the only measurement available.

Why those numbers are not a benchmark either

The obvious move would be to swap the old rule of thumb for the new range. That would be the same mistake a second time.

Sixty-three teams inside a single company are not the same thing as regulated corporate IT, a startup, or a public sector client. Approval paths, dependencies and compliance requirements differ enough that a cross-industry normal value is implausible.

So the measurement does not replace the rule of thumb. It shows something more useful: there is apparently no universal normal value for flow efficiency.

The second reason comparisons rarely hold

Beyond the make-up of the organisation sits a methodological problem: flow efficiency is not unambiguous until two things have been settled.

When does the clock start? When the request arrives, or when someone commits to working on it? There can be months of backlog between the two. Start measuring at commitment and the same organisation looks markedly better.

What counts as work? Active handling only, or coordination, reviews and testing as well? Booking waiting time inside a review step as work raises the figure without improving anything.

Both choices are defensible — but two organisations that make them differently are not measuring the same thing. Comparing across company boundaries assumes a shared definition that in practice nobody applies. Which is exactly why your own trend says more than anyone else's number.

What this means in practice

Measuring a team against 5 to 15 per cent means comparing it to a figure with no sound basis — and possibly treating a below-average flow efficiency as normal. That is the more expensive direction of the error: it stops anyone from looking.

The sensible order is the other way round. Observe your own value stream across several measurement points first — same definition, same collection, over several weeks. Then talk about changes and target values. Your own trend is the only reference that actually applies to your organisation.

And when a figure is quoted, its origin belongs with it. A metric without a source is an opinion with a decimal place.

To work out your own figure: the flow efficiency calculator puts the days from trigger to result next to the hours of actual work. Free, no sign-up, nothing leaves the browser.

Sources

• Nick Brown: Our survey says… uncovering the real numbers behind flow efficiency — 63 teams at ASOS.com, measured over twelve months. ASOS Tech Blog, 27 July 2023.
• For the widely used 5–15 per cent rule, no primary source could be established as of August 2026. Checked among others: This Is Lean (Modig/Åhlström 2012, which does not contain the ranges), David J. Anderson's spoken conference remarks, and the most widely quoted article, by Julia Wester (2016), which names no originator.