Five tests for content marketing agency benchmarks

What to take away

Why does benchmark research fail without shared definitions?

Cost per enquiry is the standard example. One agency counts paid media only. Another adds salaries, software and freelance copy. Both label the result the same way.

The gap between those two figures is often wider than the gap between strong and weak performance. A client can change supplier and appear to improve while nothing changes.

Attribution windows cause similar trouble. A 30 day window and a 90 day window will not produce the same cost per lead, so state the window beside the number.

Write a one line definition for every metric you report. If the definition needs a paragraph, the metric is not ready to appear in benchmark research. Our measurement and reporting guide sets out how to agree those definitions before a project starts.

How should you score a benchmark source?

Run each source through five tests before you quote it. The rubric below takes about ten minutes per source and catches most problems.

A total near the bottom of the range makes the source useful for context. It should not become a target in a client plan. Record the score and the reviewer's name next to the figure, so a later reader can see why it was accepted.

If you reuse a chart or a photograph from a published survey, check the permission position. The copyright notice on digital images, photographs and the internet explains when reuse is allowed and when it is not.

Where should an England benchmark set its sample boundary?

Sample boundaries matter more than headline sample size. A benchmark that pools every client size hides the difference between a two person studio and a thirty person agency.

Audience composition varies by region too. The ONS internet access reference tables give household and individual usage figures that can help assess whether a national sample reflects the people a client wants to reach; they provide audience context, not content marketing agency performance benchmarks.

Split your benchmarks by size band and sector before you compare. Where a source reports the four nations separately, keep those figures separate rather than averaging them together. Averaging across them hides which market performed best.

Weighting is the next check. A survey that over samples large agencies will drag an average upward, so ask whether the published figure is weighted to the population it describes.

How do you report a benchmark without overclaiming?

Give the comparison base first. State the range, then the source and its date. A single figure invites the reader to treat it as a standard.

Section 3 of the UK CAP Code, including rule 3.1 (misleading advertising) and rule 3.7 (substantiation), is relevant: marketing communications must not materially mislead, and objective claims need documentary evidence before publication. The ASA handles complaints and administers the Code. Keep the evidence file for a benchmark claim alongside its source.

Build the benchmark into the reporting routine rather than a one off slide. A reporting dashboard can keep the defined metrics and their sources in one place, which makes the comparison repeatable.

Date the claim in the sentence itself rather than in a footnote, because footnotes travel badly when a slide is copied.

Common questions

Does a benchmark need a minimum sample size?

There is no universal threshold. Ask whether the sample supports the level of splitting you plan, because sector and size breakdowns need more responses than a single overall figure.

Can we quote a global benchmark for an England client?

Yes, with the geography labelled and a note on cost differences. Treat it as a starting range and replace it with local data as soon as you have enough of your own.

How often should benchmark figures be refreshed?

There is no single refresh cadence for external benchmarks: check each source's publication date and stated collection window. Review your own client data each quarter.

What if two credible sources disagree?

Report both and show the range. Explain the definitional difference where you can find it, then let the reader choose which one fits their client.