Check what PPC agencies benchmark research can and cannot tell you

Scenario: deciding whether £40 CPA supports a budget increase

Suppose an agency is asked whether to increase the budget for a client trading in one English city. A published £40 cost per acquisition looks like possible support, but it is not a recommendation by itself. The decision is whether that figure can justify a budget change for this account.

Tracking definitions also sit within the law. The Data Protection Act 2018 governs how audience data gathered for paid search may be held and used. Decisions about consent and retention affect what a study can include, so record them alongside the conversion and attribution definitions.

3. Check what each source actually represents. LocaliQ, formerly WordStream, publishes annual Google Ads benchmarks by industry, including average click-through rate, cost per click, conversion rate and cost per lead. Its data is US-focused, so it can inform a range for this English client, but not serve as a UK target.

Databox Benchmarks compares connected accounts with anonymous aggregated data across Google Ads and Facebook Ads. Its sample depends on users opting in, rather than random selection. Google Analytics 4 benchmarking reports also require opt-in data sharing, cover peers in some industries, and vary in access by industry and region.

Microsoft Advertising provides industry benchmarks inside its platform, split by country and industry, including click-through rate and conversion rate. They can inform a Microsoft Ads comparison, but should not stand in for Google data.

Semrush Advertising Research and SpyFu estimate competitor paid search activity, such as keywords, ad copy and estimated spend or clicks; these are modelled figures, not audited client results.

4. Test whether the comparison travels. Scope sector, country and account size band, then ask for the sample size and collection dates. Undated figures are unusable. A UK-wide average also combines Scottish, Welsh and Northern Irish markets with English ones, so it may not describe a client trading in just one English city.

Check the account settings as well as its location. Networks, budgets, bidding choices and locations affect how ads serve and cost; Google's guidance on campaign settings sets out those controls.

Accounts with different settings are not directly comparable. Many studies also depend on agencies choosing to submit, so the sample reflects confident teams, a bias to name in a board paper.

5. State the finding and decide what to do. The source check does not establish that this client's CPA should be £40 or that a budget increase is justified. Treat the published average as a hypothesis, test it against the account, and report the gap. Show a range alongside the average because results can spread widely.

Rules to carry into the next comparison

The scenario leaves a short set of safeguards: match the evidence to the decision, record definitions and source limits, then test the average against account data. A benchmark describes how a comparable set of accounts behaves; it cannot declare what this particular account should achieve.

Date each stored figure and note exclusions, including brand campaigns or non-UK traffic. Those exclusions can explain much of the difference between an account and a published average. Use the same definitions across accounts each quarter to follow the spread, and date every figure.

Run the same checks before quoting a number to a client: request its sample, date range and currency, and record the conversion definition and attribution window in the reporting contract.

This gives later reporting a consistent yardstick. CIM's strategy and planning courses cover setting an objective before choosing a number, while the measurement and reporting guide explains how definitions carry into client reporting.

How large should a comparison sample be?

It should be large enough that a single strong or weak performer cannot shift the average. Ask how the publisher recruited participants as well as how many contributed. Research based on voluntary submissions can over-represent agencies with healthy results, so sample size alone does not settle whether a benchmark is dependable.

Is a UK average suitable for an English account?

Use it cautiously. A national figure combines markets with different costs and competition, and may not fit a client operating in one English city. Compare sector, spend band and channel mix as closely as possible before drawing a conclusion, and name any remaining difference in the report.

When should research be checked again?

Annual refreshes are usually enough for planning, alongside a quarterly check of the account's own data. Platforms change settings and bidding options, so a figure that is two years old can mislead a client about what is achievable. Keep the definitions consistent when reviewing the account over time.