Social media agency benchmarks: when to trust or check them

What to take away

A Manchester marketing lead comparing three agencies gets three different "industry averages" for engagement rate, because each agency's benchmark study samples its own client book rather than the market.
Treat any benchmark figure as a claim about a defined sample, not a fact about your sector, until you check the sample size, date and metric definition.
External data is useful context, not a substitute for your own baseline: ONS employment figures and CIM training pages describe the labour market, not your account.
For consumer-facing advertising, the ASA's CAP Code governs misleading claims; the CMA enforces consumer protection law under the Digital Markets, Competition and Consumers Act 2024.
Ask for the raw numbers behind any headline percentage before you sign a retainer built on it.

Why benchmark figures differ

Two agencies can quote the same figure, for example an average engagement rate of 4%, and both be honest. One may be averaging across every post it published last quarter. The other may be averaging only posts that beat a minimum reach threshold.

Same 4% figure, two definitions

Agency A

Post sample: All posts
Engagement: All interactions
Paid reach: Included
Quoted rate: 4%

Agency B

Post sample: Reach threshold
Engagement: Selected only
Paid reach: Excluded
Quoted rate: 4%

The metric definition changes the number more than the market does. Ask what counts as a post, what counts as engagement, and whether paid reach is included. If the answers are vague, the figure is decoration.

Ask the agency to provide its written metric definition with each benchmark figure, so you can assess whether the comparison is like for like.

Sample size and skew

A study drawn from twelve retained clients in one sector is a case study. A study drawn from several hundred accounts across sectors is closer to a benchmark, but still reflects whoever agreed to share data.

Self-selection matters. Agencies publish benchmarks when the numbers flatter them. Accounts that underperformed are quiet, which narrows what you can infer without making the data false.

For published context, look up Rival IQ's Social Media Industry Benchmark Report and Socialinsider's Social Media Benchmarks. Treat them as comparisons, not proof that an agency's client average represents the market or predicts your results.

Check whether the platform, sector, geography, reporting period, account selection, sample and engagement formula match the agency's claim. Note the report edition and exact metric; if the definitions do not match or the sample is not disclosed, the report cannot validate the pitch.

Labour market context

Benchmark reports often quote sector growth in staffing or spend. Some of it can be checked against public data rather than taken on trust.

The Employment in the UK bulletin from the Office for National Statistics gives the national labour market picture, including employment and vacancy levels. That is the context for how quickly an agency can recruit specialist social staff. It is not a social media benchmark, and no agency should present it as one.

Skills demand is visible in another way. The social and content marketing training courses listed by CIM show the discipline scope agencies are expected to cover, from content planning to paid social. A pitch that promises all of it from one junior hire deserves a question.

Rules that catch claims

For consumer-facing ads, the Advertising Standards Authority (ASA) applies the CAP Code. Its rules require objective claims to be supported by documentary evidence held before publication; you can report a misleading ad through the ASA's online complaints form. The Competition and Markets Authority (CMA) enforces consumer protection law. The Consumer Protection from Unfair Trading Regulations 2008 have been replaced, for unfair commercial practices, by the Digital Markets, Competition and Consumers Act 2024.

The safer route is a claim you can evidence with a dated source. "Our clients averaged X in Q2 2026, sample of 40 accounts, paid reach excluded" survives scrutiny. "Agencies like ours typically deliver 3x growth" does not. The measurement and reporting guide sets out fields to record so figures can be audited.

A glossary for benchmark reports

Sample: the accounts or posts a figure is calculated from.
Median: the middle value, less distorted by one viral post than a mean.
Baseline: your own pre-campaign performance, the comparison that actually matters.
Normalisation: adjusting figures so accounts of different sizes can be compared.
Attribution window: how long after a post or a click a conversion is credited.

Common questions

Should I use an agency's benchmark report in a board paper?

Use it as a labelled comparison, not as evidence of what your account will achieve. Name the source, the date and the sample, or leave it out.

What sample size makes a benchmark credible?

There is no fixed threshold, but a benchmark built on a small client book is indicative rather than reliable. Ask how accounts were selected, how many were included and the dates covered before you rely on it.

Are published benchmarks ever useful?

Yes, for spotting directional shifts and for challenging a pitch that looks out of line with the wider market. They are weakest when used to set a target.

Can a client ask for the underlying data?

You can ask, and a credible agency will share at least the methodology. If it will not, treat the headline figure as marketing copy.