Operations
Part of Running social listening operations with access, rights and recovery kept apart
Why a Brandwatch Consumer Research launch stays on hold after desk review
Desk-review a Brandwatch Consumer Research launch in England against source, classification, rights, security and recovery evidence, with a clear hold result.
What to take away
- The answer is HOLD. Brandwatch Consumer Research cannot be released for an England-focused listening service on public evidence alone.
- Five items are missing and are listed once below as the acceptance list: contract and tenant evidence, source entitlement records, controlled test results, classifier accuracy data, and a named human reviewer.
- UK GDPR erasure is conditional and the ICO page is under DUAA review, so the unseen configuration stays unresolved.
- A PASS needs measured results per gate with named sign-off, and would cover one version and one date.
- The desk holds no Brandwatch or competitor relationship that alters the verdict.
What the product record covers
Brandwatch Consumer Research is a social listening platform. Its public page lists Boolean queries, categories, automated sentiment, dashboards, alerts, exports and workflow functions. An English buyer would use it to track mentions, group them by theme and route items to a response team.
Those remain supplier claims on a global product page. They do not show which sources a contract permits, which fields are included, how complete returns are, or how well sentiment scoring matches this buyer's material.
The product was chosen because its page defines something concrete to test, not because the desk endorses it. For a wider comparison of tool claims against source questions, read five social listening tools compared on supplier claims and source questions, which explains why feature lists rarely answer entitlement questions.
What the desk did and did not test
Research closed on 5 September 2026. The desk compared one first-party product record with current UK authority guidance across ten gates.
Those gates began with source permission, privacy design, IP use, classification quality and security. Incident response, rights decisions, deletion propagation, analyst approval and supplier change control completed the ten.
The desk handled no live data, ran no queries and tested no recovery. There were no affiliate links, payments or samples, and no competing product was reviewed.
The acceptance list is defined here, once. It has five items: a signed contract with tenant evidence; per-feed source entitlement records; controlled test results on a fixed query; classifier accuracy data from an independently coded sample; and a named human reviewer with conflicts recorded. Every later reference in this review points to that list.
The UK legal test that is unresolved
Data protection is the hardest gate. The ICO states that the right to erasure is conditional and can require attention to recipients and backups. The page itself is under DUAA review.
Without a data-flow map and a documented lawful basis, nobody can decide how erasure applies here. Consent and legitimate interests both need evidence the desk never saw. No approved deletion has been traced through indexes, exports, dashboards or archives.
The operating workflow from bounded request to approved deletion matters because erasure usually fails at the last destination, not the first.
What a measurable PASS would require
PASS is not confidence after a demonstration. It is a set of recorded results.
| Gate | Measured result needed | Artefact |
|---|---|---|
| Source entitlement | One current record per feed and purpose | Contract schedule |
| Classification | Fixed query scored against a coded sample | Accuracy report |
| Rights | Erasure request traced to every destination | Test log |
| Security | Role test and log sample | Assurance pack |
| Reviewer | Named owner with conflicts recorded | Signed note |
NCSC supplier assurance questions cover ownership, incidents, data, testing and exit. They demand evidence rather than proving that any supplier meets the bar.
For classification quality, the AQuA Book separates successful execution from fitness for purpose. A dashboard screenshot establishes neither retrieval quality nor sentiment validity.
Decision and reviewer position
The outcome is HOLD, not a general fail for the product. Release can be reconsidered only after the acceptance list is complete.
Launch decision path
Does mandatory evidence pass?
Reconsider release for that version and date
Keep launch on HOLD
- Signed contract and contracted edition on file.
- One entitlement record per feed and purpose.
- Controlled test results with query version and date.
- Deletion traced through every destination.
- Named human reviewer with conflicts recorded.
Before resubmission, work the gates in the social listening supplier selection checklist, which puts source permission ahead of feature scoring.
Qualified privacy, IP, security, accessibility, research-quality and commercial reviewers must sign the same configuration. Even a pass covers that version and date only, not guaranteed compliance, full coverage, classifier accuracy or business results.
On relationships, the desk has no affiliate arrangement, payment, sample or commercial tie to Brandwatch, and none to competing listening vendors. A named human reviewer is still unassigned, and that gap is itself a failed gate.
Common questions
How accurate is the sentiment classification?
Unknown. No fixed query and no independently coded sample were supplied. Treat any accuracy figure on a public page as a supplier claim until it is reproduced on your own material.
Which sources is an English buyer entitled to use?
That depends on the signed contract, not the product page. Each feed needs one current entitlement record tied to a stated purpose, because platform terms change and content can be removed.
What would change the verdict?
Completion of the acceptance list, with measured results per gate and a named reviewer whose conflicts are recorded. Only then can a launch owner ask for release.
