Operations
Part of Operating retail media so a campaign record survives every handover
A retail media placement quality record that logs untested conditions beside tested ones
Retail media placement record sets out evidence states for tested and untested conditions, plus permission, presentation and measurement questions before a decision.
A retail media quality checklist should record, for one named placement, what was tested and what was not, so an untested permission or presentation condition stays visible rather than reading as a pass. Use one record per placement instead of awarding a platform a general quality badge.
This checklist was prepared from public sources on 6 September 2026. Every check is proposed work unless a separate evidence record shows it was performed. No retailer implementation, campaign or shopper panel was tested.
What to take away
- One placement gets one recordproperty, placement, creative version and environment.
- Untested conditions stay labelled untested, never converted into assumed passes.
- A screenshot documents what a reviewer saw, not whether shoppers noticed the label.
- A credited sale stays labelled attributed, because causation needs a counterfactual.
- Permission and presentation gaps outrank measurement gaps in the handover.
Identify the assessment object first
Enter the property, placement, creative version and environment, plus the advertiser's approved destination and its catalogue record. State which page conditions are in scope, so a second reviewer can tell the assessed placement apart from a similar one elsewhere. The entry gates for one England retailer explain why the object is named before inventory is sold.
These evidence states are house labels, not standardised grades.
| Evidence state | What it records | Next step |
|---|---|---|
| Observed | Reviewer saw the state and kept the rendered result | Attach the artefact to the record |
| Unresolved | Evidence was requested but has not arrived | Name the owner and the open question |
| Not tested | Condition was left outside this review | Record why it was excluded |
| Excluded | Condition sits outside the decision in question | Note who holds that evidence |
Example of a completed entry
Field / Entry
- Placement
- Sponsored banner, grocery retailer recipe page, desktop
- Creative version
- Approved version 3, compared with the live render
- Evidence state
- Rendering observed, consent handling unresolved
- Unresolved item
- Which tag fires on the impression and what it writes to the device
- Owner
- Agency tag manager
- Evidence requested
- Tag list naming each recipient supplier and its retention period
Inspect the shopper-facing presentation
Section 2 of the CAP Code carries the advertising recognition requirement. Record how the placement is identified as advertising in its page context, then compare the rendered creative and destination with the authorised versions. Note any state where labels, content or layout differ, and preserve the rendered result.
A screenshot can document what a reviewer saw in that state. It cannot show that shoppers noticed the label or understood the offer. Those claims need observed user evidence with a disclosed method, and the advertising rules checks cover the booking side of the same question.
Shopper-Facing Presentation Inspection
- Record how placement is identified as advertising
- Compare creative and destination with authorised versions
- Describe states where labels, content or layout differ
- Preserve evidence of rendered result
- Identify excluded conditions
- Note screenshot limitscannot show shopper notice
Examine the associated data behaviour
List the data flows before a privacy specialist sees the build: which device or browser identifiers are read, which tags fire on which event, which suppliers receive each event, the stated purpose and the retention period. How do the rules apply to online advertising? covers consent for storage and access technologies.
Request evidence of withdrawal handling: where a withdrawal is recorded, how quickly it stops the tag or the processing, and whether a suppression list reaches every recipient. A consent banner answers none of those questions. The ICO's UK GDPR guide sets out the records a controller must be able to produce.
If an integration cannot be inspected, record the limit and who must obtain further evidence.
Data Behaviour Review
- Refer implementation to competent privacy reviewer
- Provide actual device and supplier data flows
- Identify permission conditions assessed
- Identify evidence for withdrawal handling
- Do not infer compliance from consent banner
- Record limits and who must obtain further evidence
Check what each measurement represents
The IAB Europe Commerce Media Measurement Standards V2.1 (May 2026) separates impression concepts from viewability and addresses attribution, so keep the adopted specification beside the inspection record. Ask where an ad request is counted, what establishes rendering, and how visibility is collected, then document filtering without inventing a delivery rate.
Keep a credited sale labelled attributed, because a causal claim needs evidence about what would have happened without the advertising. Click-based and view-based attribution compared is worth reading before a report is accepted as proof.
Prioritise unresolved items before handover
- Permission gaps first, since no presentation or measurement fix compensates for an open consent question.
- Presentation gaps second, because recognition and creative accuracy affect shoppers directly.
- Measurement gaps third, because definitions can be corrected in a later reporting cycle.
For each open item, state the affected scope, the evidence needed and the owner. Avoid a numerical score, because completed unrelated checks can hide an unanswered permission issue. Send the release owner the whole record, including exclusions.
Common questions
What should a retail media quality checklist be used for?
It records evidence for a particular placement, not a general quality badge. An England retailer needs to know which version was examined, what the reviewer observed and what remains untested.
Can a screenshot prove that shoppers understood an advertising label?
No. It documents what a reviewer saw in that state, but not that shoppers noticed the label. Those claims need observed user evidence with a disclosed method.
Does an attribution report prove that advertising caused a sale?
No. A credited sale stays labelled attributed, and causation needs evidence about what would have happened without the advertising.
Who should receive the finding record?
The release owner receives the whole record, including exclusions. A revised build is linked to the new observation, and earlier findings are not erased.
