Strategy
Part of A retail media plan for 2027 with evidence gates, named owners and practical stop rules
On-site, off-site or in-store: comparing retail media routes for one England case
Compare on-site, off-site, in-store and current retail-media routes for one England case using common evidence, cost, control and exit fields.
For this comparison, channel means where the advert is delivered: the retailer's digital property, external inventory or a physical retail environment. It does not mean acquisition source, commercial model or measurement vendor. A buyer should compare every route against one approved product, customer task, advertiser exchange, England context and period.
No provider or product is named, tested or recommended. Research was completed on 6 September 2026 from the cited public records. Costs, audience, availability, performance and buyer thresholds remain unknown.
The common case
Assume an England retailer is considering one paid placement for an approved product. The advertiser has supplied a factual claim file and creative. The customer should recognise the advert, see accurate product and price information, retain a usable privacy choice and complete the same accessible shopping task. Only synthetic or specifically authorised events may enter the comparison test.
IAB UK's category page divides retail media into on-site, off-site and in-store environments. Its off-site definition describes retailer data used to buy inventory outside the retailer's shopping platforms, sometimes through agencies or platforms. These are industry descriptions, not evidence of rights, availability or results for this case.
Compare identical fields
| Field | Current route | On-site | Off-site | In-store |
|---|---|---|---|---|
| Inventory and presentation owner | record | record | record | record |
| Customer task and eligible opportunity | same | same | same | same |
| Ad-recognition fixture | required | required | required | required |
| Data sources, recipients and device operations | map | map | map | map |
| Accessible task and alternatives | test | test | test | test |
| Exposure and outcome definitions | same unit | same unit | same unit | same unit |
| Cash and economic cost | blank | blank | blank | blank |
| Correction, withdrawal and incident route | demonstrate | demonstrate | demonstrate | demonstrate |
| Export, retention and exit | demonstrate | demonstrate | demonstrate | demonstrate |
The current route is not assumed to be free or safe. Record its labour, defects, missing data and contractual constraints in the same units.
Differences that matter
On-site delivery may keep presentation within a retailer-controlled website or app, but the actual serving, measurement and data recipients still need mapping. Off-site delivery adds external inventory and can alter the controller, publisher, creative and tracking chain. In-store delivery introduces physical placement, staff operations, screen or audio access, and a route for customers who cannot use the main presentation.
CAP Code rule 2.1 requires marketing communications to be obviously identifiable in its section on ad recognition. Apply that test separately to each final context. A label that is conspicuous on a product page may fail when resized, voiced, cropped or shown on a store screen.
The final ICO storage and access technologies guidance covers cookies, tracking pixels, fingerprinting and comparable operations under PECR and, where relevant, UK GDPR. Off-site reach does not excuse the buyer from recording each operation, purpose, recipient and refusal path.
Measurement cannot choose the route alone
IAB Europe published Commerce Media Measurement Standards V2.1 on 22 January 2026. They provide industry metric definitions for their stated scope. Use one definition version, eligible population, period, timezone and post-order treatment across channels. Mark unsupported mappings as unavailable instead of forcing unlike events into one table.
A delivery count, click or attributed sale is not a causal benefit. Keep order, cancellation, return and refund states visible. Have a statistician approve any comparison design before interpreting a difference.
Decision
There is no overall winner. Reject a route if advertising, consumer, privacy, accessibility, security, IP, contract or exit evidence remains unresolved. Among routes that pass every gate, the buyer can apply blank, predeclared weights to control, cost and measurement evidence. Re-open the records when inventory, data recipients, creative, supplier terms or customer journeys change.