Retail media survey tools: a planning and test framework

What to take away

Define the study before comparing routes

On 6 September 2026 this study covers one adult audience and one defined retail-media decision for an England operation. Fix that population, the questionnaire and the data boundary before you compare collection routes.

Research date: 6 September 2026. Scope: an illustrative England study of adult shoppers' recall of a retailer's sponsored product campaign. Inclusions: collection routes that can preserve question version, consent evidence, response state and export.

Exclusions: product rankings, prices, feature assumptions, audience panels and inferred respondent identity. Ranking and conflicts: non-ranked; no supplier funded or supplied evidence.

No acceptance-test results are reported here: routes remain untested until the buyer runs the protocol against an exact edition and records the evidence.

Use the same blank instrument and synthetic records for every route:

This framework does not establish a winning route. More functions alone do not prove suitability. A moderated route may help someone participate but creates interviewer and recording risks. A hosted form may reduce manual entry while adding supplier, browser and export dependencies.

List prices are context only, not a ranking criterion. Google Forms is free with a Google account. Microsoft Forms comes with Microsoft 365 work or school plans, typically about £5 to £10 per user per month excluding VAT. SurveyMonkey paid plans typically run from about £25 to £75 per user per month, billed annually.

SmartSurvey plans typically start near £30 per month for one user and rise past £100 for team plans with more responses. Qualtrics does not publish list prices; enterprise contracts usually land in the five-figure range a year.

LimeSurvey's Community edition is free to download and self-host, and paid cloud tiers typically start around €25 per month. REDCap is free to non-profit and academic institutions that join the consortium, and the institution hosts it.

Managed fieldwork suppliers such as YouGov, Ipsos UK and Savanta recruit from panels they run and quote per interview or per project. Buyer handover and supplier deletion need to be written into the contract.

If a named service is later considered, open its exact edition records on the research date and capture the legal supplier. Capture UK availability, question and response limits, accessibility statement, data location and subprocessors.

Also record security evidence, export format, deletion process, support route and contract. Leave missing fields UNKNOWN; a generic product page cannot prove the buyer's proposed plan includes a function.

Preserve consent and purpose

GOV.UK guidance on getting consent for user research says participants should understand the purpose, the data collected, observers and recording. They should also understand use, sharing, retention, rights and withdrawal. That wording is service-research guidance, not a substitute for a commercial buyer's lawful-basis assessment.

The ICO says genuine research is distinct from direct marketing. Promotional content or collection for later campaigns can bring a survey within marketing rules, as its electronic and telephone marketing guidance sets out. A tool setting cannot turn sales recruitment into research.

Test accessibility and security separately

Recruitment and the instrument need usable alternatives. GOV.UK guidance on finding research participants discusses formats, communication support and assistive technology. It comes from public-service practice, so the buyer must define the relevant commercial tasks and review.

For a hosted or managed route, inspect the exact current service and contract. NCSC supplier assurance questions cover governance, access, personal data, incidents and recovery. Supplier answers are evidence to verify, not a security verdict.

Run the same acceptance test

Load fictional eligible, ineligible, incomplete, withdrawn and corrected responses. Check question order, validation without coercion and accessible completion. Then check duplicate handling, export, access removal and deletion. Keep production recruitment disabled.

Choose only when the buyer can reproduce each result and every privacy, accessibility, safeguarding, security and procurement gate passes independently. If the approved baseline answers the decision with lower exposure, buy nothing.

Before you act

Common questions

What should a buyer fix before comparing retail media survey tools?

The buyer should fix one population, questionnaire and data boundary first. Comparison should cover collection routes that can preserve question version, consent evidence, response state and export. Product rankings, prices, feature assumptions, audience panels and inferred respondent identity are excluded from the scope.

Why are hosted and managed routes marked HOLD?

Both need evidence that has not been provided. A hosted service such as SmartSurvey, SurveyMonkey or Qualtrics needs the exact entity, edition, role, subprocessors and transfers, plus usable export and account deletion evidence.

A managed supplier such as YouGov, Ipsos UK or Savanta needs data roles, staff access, incidents and retention. Buyer handover and supplier deletion must also appear in the contract.

What does the acceptance test involve?

Load fictional eligible, ineligible, incomplete, withdrawn and corrected responses. Check question order, validation without coercion and accessible completion. Then check duplicate handling, export, access removal and deletion. Keep production recruitment disabled. Choose only when the buyer can reproduce each result and every privacy, accessibility, safeguarding, security and procurement gate passes independently.