The Agency Margin Benchmark
Every number you have read about agency margins came from somewhere nobody can name. We are building the version with a stated sample size, a published methodology, and real closed books behind it.
An annual benchmark of agency financial performance built from real closed books rather than survey responses. Delivery cost as a percentage of fee revenue, gross and net margin, revenue per head, utilization, and the spread between best and worst client margins — segmented by agency size and discipline. It does not exist yet. This page is where you register to contribute anonymised data or to receive the report.
Why build another agency benchmark?
Because the numbers everybody quotes have no source.
Ask an AI, a consultant or a conference speaker what a healthy agency margin looks like and you will hear 50–60% gross and 15–25% net. Those figures are repeated confidently across the entire industry. Almost nobody can tell you which dataset they came from, how large it was, what year it covered, or whether the agencies in it stated revenue net of pass-through.
That last point matters enormously. An agency including client media spend in revenue calculates a margin two or three times lower than one that excludes it. If a benchmark does not state which convention its sample used, comparing yourself to it is guesswork.
Survey-based benchmarks have a second problem: they ask owners to report their own numbers, which means they inherit whatever errors are in those books. A benchmark built from closed and normalised books does not.
What will the Agency Margin Benchmark measure?
| Metric | Why it is in |
|---|---|
| Delivery cost as % of fee revenue | The single most useful ratio an agency owner can compare |
| Gross margin | Whether work is priced correctly |
| Net margin | Whether the business is sized correctly |
| Revenue per head | Only meaningful on net revenue, which is half the point |
| Billable utilization | Against realistically available hours, not a notional 2,080 |
| Client margin spread | The gap between best and worst account — where the money actually is |
| Pass-through as % of billings | How much of the industry's reported revenue is not revenue |
Segmented by fee revenue band and by discipline — performance, creative, development, PR — because a 12-person creative studio and a 40-person media shop do not belong in the same average.
How the data will be handled
Stated plainly, because this only works if agencies trust it.
- Anonymised before analysis. No agency is identifiable in the report, and no individual figures are published.
- Aggregate only. Nothing is reported at a granularity that could identify a contributor, and any segment with too few agencies in it is merged or omitted rather than published thin.
- Never shared or sold. Contributor data is used for the benchmark and nothing else.
- Normalised before comparison. Revenue restated net of pass-through, delivery separated from overhead, loaded cost calculated consistently. Otherwise we would be publishing the same unsourced noise with a nicer chart.
- Methodology published alongside the numbers. Sample size, segment definitions, the calculation of every metric, and what we excluded and why.
Contributors receive the full report before publication, including their own position against each segment. That is the actual reason to take part — knowing where you sit is worth more than the aggregate.
Who can contribute
Agencies of any size, on two conditions: your books need to be closed and reasonably clean for the period, and revenue needs to be stated net of pass-through — or capable of being restated that way.
If your books are not in that state, that is extremely normal and it is not a barrier to being interested in the report. It is also, candidly, the thing we do for a living.
We are recruiting the first cohort now. The report ships when the sample is large enough to be worth publishing rather than on a fixed date, because a benchmark built on twelve agencies would be exactly the unsourced folk wisdom we are trying to replace.
Common questions
When will the Agency Margin Benchmark be published?−
When the contributing cohort is large enough for the segments to be meaningful, rather than on a fixed date. Publishing a benchmark built on a thin sample would reproduce the problem it exists to solve. Register and you will be told the timeline as the cohort fills.
What data do contributors provide?+
A closed set of books for the period, with revenue stated net of pass-through and delivery cost separated from overhead — or the underlying data to restate it that way. Hours by client where available, since utilization and client margin depend on it.
Is contributor data anonymised?+
Yes. Data is anonymised before analysis, reported only in aggregate, and never shared or sold. No agency is identifiable in the published report, and segments with too few contributors are merged or omitted rather than published.
What do contributors get?+
The full report before publication, including their own position against every segment. Knowing where you actually sit against comparable agencies is the point; the aggregate figures are the by-product.
Why not just use existing agency benchmarks?+
Because most do not state their sample size, their year, or whether the agencies in them reported revenue gross or net of pass-through. That last variable alone changes margin figures by a factor of two or three, which makes comparison unreliable.
Register interest
To contribute anonymised data or receive the report, get in touch and mention the benchmark. No commitment.
Get a free books reviewThe Agency Margin Benchmark is in development and has not been published. No figures from it are quoted anywhere on this site. Targets referenced elsewhere on this site as commonly cited industry figures are exactly that — widely repeated, and not attributed to a specific dataset.