Reviewing retail media products: who did the work, what was frozen and what was tested

A retail media review methodology should define its subject and evidential limits before anyone evaluates a provider. For England retailers, start with the advertising inventory and related data operation under consideration. A product brochure, an implemented placement and an agency engagement require different assessment methods.

Record who commissioned and performed the work

The intended publication is shopperads.co.uk. This method was prepared through AI-assisted desk research on 6 September 2026; no product-testing body or named human reviewer has been appointed. Independent fact-checking and qualified advertising, privacy and security review remain pending.

Publisher sponsorships, affiliate arrangements and supplier relationships have not been verified. No paid account, sample, demonstration or supplier contact was used for this research. Confirm the wider conflicts position before publication rather than treating absence of purchased access as proof of independence.

Freeze a precise subject

For the documentary example, the subject is Kevel's Retail Media Guide as inspected on 6 September 2026. A publication version was not established. No corresponding retailer, software build or deployed configuration was supplied.

The guide describes sponsored listings and separates returned ad content from rendering by the implementing property. That global first-party record supports questions about responsibility boundaries. It does not establish that a retailer's placement rendered correctly or that the product was tested by this publication.

For a later operational assessment, record the exact build, environment and authorised task. Describe the retailer property and placement scope so another reviewer can distinguish the subject from a different implementation using the same supplier.

Decide what each evidence type permits

Classify supplier descriptions, demonstrations, direct observations and independent studies separately. Record their origins, inspection dates and limitations. External guidance supplies assessment criteria; it is not an independent test result for the product under review.

The NCSC's supply-chain security guidance addresses dependencies and access to data. It supports requesting evidence about the actual service chain. It neither certifies Kevel nor prescribes this editorial review procedure.

Define the questions the proposed assessment will answer and the records needed for each. Where evidence is unavailable, use an explicit unresolved state. Do not allow a broad favourable summary to obscure an unexamined security or privacy question.

Specify tests before claiming results

If hands-on work is commissioned later, document the authorised conditions, method, observer and evidence capture. Identify what was excluded and why. Supplier-selected examples should remain labelled as such rather than being represented as independent sampling.

No integration, campaign, support response, screenshot, accessibility check or user study was examined here. Performance, usability, reliability and customer satisfaction therefore have no observed basis in this article. No scores are calculated.

Keep requests, returned responses, rendering and viewability separate in any future evidence plan. Attention needs its own method. Attributed sales do not establish causal incrementality, and a modelled result requires disclosed assumptions and limitations.

Review the wording of the finding

CAP's rules on substantiation concern objective claims in marketing communications. They are a UK advertising-code source, not a blanket classification of all review writing as advertising. Refer the actual publication context to a competent reviewer.

The proposed editorial check is to trace each evaluative sentence to evidence of matching scope. A statement that documentation describes a feature must not become an assertion that the feature works reliably in England deployments.

Close with the next evidence requirement

This documentary method identifies how a review could be commissioned; it offers no product endorsement or deployment verdict. Appoint the reviewing body, disclose conflicts and obtain the specific implementation evidence before evaluating outcomes. Preserve source and interpretation changes in a correction record, with refreshed references and genuine named review required for publication.