An evidence ledger for retail media competitors that separates facts from supplier claims

Retail media competitor research should show what a named source establishes and what remains unknown. It cannot infer England market share, advertiser demand, audience reach, performance or customer satisfaction from search visibility, a company filing or a supplier page.

Fix the comparison unit

• Define one retailer-advertiser job and the on-site, off-site or in-store environment.
• List the role being compared: retailer media owner, network, agency, platform, data supplier or measurement provider.
• Record the research cut-off, England connection and exclusion rules.
• Keep the current internal or no-purchase route in the evidence table.

IAB UK's retail media definition supplies a category boundary. It does not establish that a business offers every environment or owns the inventory it describes.

Separate four evidence classes

1. Register fact: legal name, number, status, filing and registered office exactly as recorded.
2. First-party claim: the supplier's own description of its current service, geography, client requirement or method.
3. Independent evidence: an authoritative decision, standard, dataset or buyer-controlled test with a stated population and date.
4. Unknown: anything not supported for the exact entity, service and period.

Companies House explains that its public register contains company information and makes certain records public. A filing can identify an entity; it cannot prove retail-media capability or customer outcomes. Record brands separately from legal contracting parties.

Inspect claims without copying a database

• Save the URL, page title, access date, exact claim and stated geography.
• Distinguish current availability from a case study, announcement or expired page.
• Check whether the claim concerns advertising inventory, data, technology, managed work or measurement.
• Record price, reach, return, client and performance fields as UNKNOWN unless direct compatible evidence exists.
• Respect site terms, access controls, copyright and database rights.

The Intellectual Property Office's database-right guidance explains that database contents and original selection or arrangement can have separate protection. Public access is not blanket permission to reproduce a competitor's collection.

Treat reviews as a distinct source

• Keep supplier testimonials, platform reviews, regulator cases and buyer interviews in separate rows.
• Verify quotation permission and context before any participant words are published.
• Do not calculate sentiment or popularity from an undisclosed or self-selected sample.
• Record open investigations as open, not as findings.

The CMA's online consumer reviews case record documents current enforcement and guidance activity. It does not establish the quality of every named retail-media supplier. A retailer's customer review also concerns the retail transaction unless the evidence directly addresses the media service.

Minimise people and security data

• Collect business-contact information only for a defined purpose and approved route.
• Exclude home addresses, private profiles, access credentials, customer records and inferred protected traits.
• Restrict the ledger, record corrections and delete data under the approved schedule.
• Do not probe live accounts, tracking, adverts or endpoints to complete a desk comparison.

Under the ICO's data minimisation guidance, a project should restrict personal information to what is adequate, relevant and necessary for its defined purpose. NCSC supplier assurance questions can structure authorised evidence requests; they do not authorise security testing.

Finish with an evidence-gap table, not a league table. If two routes cannot be compared on the same inventory, commercial unit, data role, measurement definition and exit evidence, report them as non-comparable. Keep the project on hold until the buyer has a legitimate decision and enough compatible evidence to proceed.