Reviews
Part of Retail media reviews: a method for keeping evidence apart
Retail media review methodology: freeze the subject, classify four evidence types, label paid content
A retail media review method for England: freeze the subject, classify four evidence types, and apply DMCC Act 2024 rules on fake reviews and paid promotion.
A retail media product review is credible only when the subject is fixed before anyone judges it. That means a named build, environment, placement scope and inspection date. This method suits England retailers and publishers who want a review that a second reviewer could repeat.
What to take away
- Review an artefact and its publication, not a brandfreeze build, environment, placement scope and date first.
- Keep four evidence types apartsupplier descriptions, demonstrations, direct observation and independent studies.
- Say who did the work and what was never verified, including any commercial relationship.
- Never publish a fake review, hide a paid promotion or show reviews in a misleading way.
- Name the people who sign off before publication, and keep a correction record.
Start with the subject, not the vendor
A brochure, a live placement and an agency engagement each need a different assessment method. Begin with the advertising rules that apply to the inventory and the data operation behind it. Then write down what is actually in scope.
Use a selection checklist to record what was examined. Anything not examined stays unknown. A review that lists its gaps is more useful than one that scores them away.
Freeze the subject in a dated record
Record the exact build, the environment and the authorised task. Describe the retailer property and the placement, so a second reviewer can tell this subject from a different implementation by the same supplier.
Example: a frozen subject record
| Field | What to record | Illustrative entry |
|---|---|---|
| Artefact | Named product or module | A sponsored product listing module |
| Version or build | Build number and release date | The label shown in the admin console |
| Environment | Test tenant or live retailer property | Live property, signed-in test account |
| Placement scope | Named retailer, page and ad slot | Search results page, first sponsored slot |
| Inspection date | Date of each session | Each session logged separately |
| Reviewer | Appointed reviewer plus observer | Named reviewer plus named observer |
If a field cannot be filled, write "not established". Do not infer it from marketing material.
Separate the four evidence types
Classify supplier descriptions, demonstrations, direct observations and independent benchmark research apart from one another. Record the origin, inspection date and limits of each. Each type supports a different claim.
A supplier document that describes a feature is not evidence that the feature works on an England deployment.
The table sets out what each type can and cannot support.
| Evidence type | Can support | Cannot support |
|---|---|---|
| Supplier description | A statement about what the supplier says | That the feature works on your deployment |
| Demonstration | What the supplier chose to show | Behaviour outside the demonstration |
| Direct observation | A claim about the conditions you tested | Performance under other conditions |
| Independent study | A wider claim, within its own scope | Anything outside that scope |
Keep requests, returned responses, rendering and key metrics separate in the evidence plan. Attention needs its own method. Sales credited by attribution methods do not establish causal incrementality.
Say who did the work and what was not verified
State who carried out the review, what access they had and what was never checked. Disclose any sponsorship, affiliate arrangement or supplier relationship. Absence of purchased access is not proof of independence, so confirm the conflicts position in writing.
A review published to sell something is marketing. Under DMCC Act 2024 Schedule 20, using editorial content to promote a product where the trader has paid is a banned practice unless that is made clear. Do not publish fake consumer reviews, meaning reviews that purport to be based on genuine experience but are not.
The hidden ads principles cover the CMA, which can issue fines and directions directly under the DMCC Act, and the Advertising Codes enforced by the Advertising Standards Authority (ASA). Content creators must label content as advertising where they have been paid or otherwise incentivised to endorse a product. Apply the same test to your own review pages.
Specify tests before you claim results
Document the authorised conditions, method, observer and evidence capture before hands-on work is commissioned. Say what was excluded, and why. Supplier-chosen examples stay labelled as supplier-chosen.
Compare suppliers with the same method each time. The reviews and comparison methods article keeps documentation and test results in separate columns. A claim that rests on documentation alone must not be reworded as a reliability claim.
Steps to a review that survives scrutiny
- Appoint the reviewing body and the named observer, then publish their conflicts.
- Freeze the subject recordbuild, environment, retailer property, placement scope and inspection date.
- Classify every evidence type with its origin, date and limits.
- Specify authorised test conditions and the capture method before claiming results.
- Trace each evaluative sentence to evidence of matching scope.
- Label any paid or incentivised content, and keep a correction record for later changes.
