How to audit a marketing agency's competitor research

What to take away

Does the agency define the competitor set before it starts?

A vague brief produces vague rivals. Ask the agency to write down which competitors it will track, why each one qualifies, and which it has deliberately excluded. The definition should tie to your category, your geography and your buyer, not to whoever ranks first.

If the agency cannot explain its exclusions, the research will drift towards whichever brands are easiest to find. That is a sign of convenience, not method.

Which evidence sources will it use?

Competitor research rests on three evidence types: public filings, published marketing output, and primary interviews with buyers. Each has a different reliability level, and a good agency labels which is which.

For market sizing, point the agency to official statistics rather than blog estimates. The UK National Accounts, The Blue Book time series from the Office for National Statistics includes detailed economic statistics covering the advertising industry's output.

For audience evidence, the method matters more than the sample size. Our guide to audience research for marketing agencies explains how to judge sampling, recruitment and question design before you accept any findings.

How will it collect and handle data?

Ask where respondent data is stored, how long it is kept, and who inside the agency can see it. Under UK GDPR, the agency needs a lawful basis for processing and a retention rule it can actually follow.

If the work involves automated analysis, transcription tools or synthetic personas, check the supplier against the ICO guidance on AI and data protection. It sets out how to assess fairness, transparency and accuracy when AI tools touch personal data.

The answers should be written down. Verbal assurances are difficult to audit later.

What should the output look like?

A useful deliverable separates observation from interpretation. Observations are what the competitor published or said. Interpretation is what it might mean for you. Blending the two hides weak evidence behind confident prose.

Ask for a source column beside every claim, a confidence rating, and a short list of what the research could not establish. Honest gaps are more valuable than invented certainty.

Which skills should the research team hold?

Competitor research is a craft, not a task to hand to whoever is free. Ask who will run the interviews, who will code the responses, and what training they have completed.

The Institute of Practitioners in Advertising maintains a public list of courses and qualifications for agency staff, which gives you a benchmark for judging claimed expertise.

If the agency plans to run its own surveys, the tooling choices affect quality. Our comparison of survey tools used by marketing agencies covers panel sourcing, question logic and reporting limits.

Which tools and frameworks should the agency use?

Ask which platforms the team uses for search and traffic data. Agencies commonly work with SEMrush, Ahrefs, Similarweb and SpyFu for organic visibility, paid search and audience overlap. For ongoing monitoring, Crayon, Klue and Meltwater are widely used to store competitor profiles, alerts and battlecards.

Analysis frameworks matter as much as the software. A credible method will apply SWOT, Porter's Five Forces, win/loss analysis or perceptual mapping, and say which one it used. Agencies also mine public marketing output: Meta Ad Library and Google Ads Transparency Center show active ads, while Companies House filings give financial and ownership context.

For audience and spend evidence, UK agencies may use Nielsen, BARB, RAJAR, IPA Bellwether or Touchpoints, depending on the sector. Deliverables often take the form of battlecards, share-of-voice dashboards or win/loss reports. Ask to see a redacted example before you commission the work.

How often will the research be refreshed?

Competitor research decays. Pricing pages change, campaigns launch and positioning shifts. Agree a review rhythm at the start, usually quarterly for active categories and twice yearly for stable ones.

Each refresh should record what changed since the last version. A dated change log turns a one-off report into a trackable asset.

Comparison table: weak versus strong practice

Common questions

What is a reasonable competitor set size?

Most briefs work with five to eight direct competitors plus two or three adjacent brands that compete for the same budget. Fewer than five rarely shows a pattern. More than twelve usually dilutes the analysis without adding insight.

Should the agency interview competitors' customers?

Only where recruitment is ethical and disclosed. Interviewing your own buyers about rival brands is usually more reliable, because you can verify consent and reuse the data under your own privacy notice.

How do we test the research quality before committing?

Ask for one completed competitor profile on a brand you already know well. Compare its claims against the public record. Errors there predict errors everywhere else in the programme.

Does competitor research need a separate budget line?

Yes. If it sits inside general strategy hours, it gets cut first when deadlines tighten. A named budget line makes the refresh cycle enforceable and the output auditable.

Which tools should the agency use for competitor research?

Common choices include SEMrush, Ahrefs, Similarweb and SpyFu for search and traffic data, Crayon, Klue and Meltwater for monitoring, and Meta Ad Library or Google Ads Transparency Center for live adverts. Ask why those tools suit your category.

What frameworks should the agency apply?

Expect a named framework such as SWOT, Porter's Five Forces, win/loss analysis or perceptual mapping. The framework should appear in the proposal, not added after the data is collected.

What deliverables should we request?

Ask for battlecards, a share-of-voice dashboard or a win/loss report. Each should include sources, confidence ratings and a dated change log.

How do we check the agency's data sources?

Request the source behind every claim. Public filings, published marketing output and primary interviews should be labelled separately, with official statistics such as the ONS Blue Book used for market sizing.

What should the agency do about AI tools?

The agency should document how AI tools touch personal data and check them against ICO guidance. Fairness, transparency and accuracy must be assessed before use.

How often should the agency refresh the research?

Agree a rhythm at the start: usually quarterly for active categories and twice yearly for stable ones. Each refresh should record what changed since the last version.