Audience research
Part of Reputation management audience research: separate the populations
Researching reputation competitors without inventing their customers or traffic
Research reputation-management competitors through dated public evidence, narrow claims and explicit unknowns without inventing customers, traffic or market share.
What to take away
- Answer one named buyer question with dated public recordsentity, service, URL, capture date, geography.
- A review count or star rating is a platform metric, not a customer, traffic or market-share figure.
- Label every line as a supplier claim or an independent observation, with its source page.
- Stop when identity is unclear, collection looks disproportionate or evidence is stale.
- Give each finding an owner and a recheck trigger, and let the conclusion stay unresolved.
Fix the subject before you compare anything
Record the legal entity, trading name, exact service or edition, public URL, evidence capture date and claimed geography. A similar name is not proof that the same supplier runs a site or sells into England.
Companies House holds filed entity details, and its register search warns that filed information is not checked for accuracy. Treat incorporation and filings as narrow register facts, not proof of active service or quality.
Separate supplier claims from independent observations
Copy the exact wording and page date for each claimed function, then label it a supplier claim. Note separately only what a public page showed on a stated date, such as a visible correction route.
First-Party Claims vs Observations
Supplier claim
- Source
- Supplier wording
- Evidence
- Exact wording and page date
- Label
- Supplier claim
- Limit
- No customer outcomes
Independent observation
- Source
- Researcher check
- Evidence
- Visible on particular date
- Label
- Observation
- Limit
- No customer outcomes
The gap between a published claim and a tested service is the subject of reputation agency reviews, which is worth reading before you score any provider.
Archive pages where permitted, because wording changes without notice. Write unknown when a plan, contractual limit or data role is not public. Do not fill that gap from comparison blogs.
Why a review count is not a customer count
A review total counts published reviews on one profile. It does not count buyers, sessions or share of any market, and it shifts when reviews are removed, merged or filtered.
A star average summarises only the reviews shown. It says nothing about service quality or whether those reviews are representative, and it cannot be converted into a customer, traffic or market-share figure.
Read the policy instead: who may submit, how incentives and negative reviews are handled, and how challenges work. The CMA's guidance for online review sites sets out current UK consumer-law expectations for publishers of review information.
Do not contact reviewers for research without an authorised recruitment and privacy route.
Advertising continuity from claim to landing page
Follow a material claim from advert or search result through to the landing page and its qualifications. If a testimonial appears, record its placement and the claim it makes, not the reviewer's identity.
CAP's testimonials and endorsements advice says marketers need documentary evidence that a testimonial is genuine, and must not pose as consumers. That check does not decide whether one advert is lawful. The five checks in review-led advertising rules give a contested claim somewhere to sit.
Personal data, copyright and database limits
Public availability is not permission to profile reviewers, staff or complainants. Record the purpose, the fields you need, access, retention and deletion. The ICO's data minimisation guidance ties data volume to the stated purpose.
Keep allegations, health details and political views out of a general comparison sheet. If special-category data appears incidentally, restrict access and take qualified advice.
Record a page reference and a short factual note rather than copying a competitor's review corpus, photographs or reports. Check site terms and applicable rights before automated collection.
Example: one completed record row
Illustrative entries only; replace each one with your own captures.
One completed record row
| Field | Example entry | Note |
|---|---|---|
| Claim | "We reply to every review within one working day" | Copied from pricing page |
| Evidence type | Supplier claim, first-party | Not independent |
| Capture date | 12 February 2025 | Illustrative date |
| Scope | Public UK pages, one service tier | Not the whole company |
| Limitation | No published response-time data | Outcome unverified |
| Owner | Research lead | Named in the record |
| Recheck trigger | Policy page change or next quarter | Diary the date |
When to stop and how the record ends
Stop if identity is uncertain, collection would be disproportionate, evidence is stale, or the comparison crosses unlike service boundaries. Mark the row unresolved and say which fact would settle it.
An unresolved row beats a ranking built from incomplete public evidence. The same discipline applies to growth claims, as reputation market outlook argues.
Common questions
Can a competitor's review count tell me how many customers they have?
No. It counts published reviews on that profile on the day you read it, and moves when reviews are removed, merged or filtered. It cannot become a customer or market-share number.
Why keep supplier claims and independent observations apart?
A claim records what a supplier says it does. An observation records what a public page showed on a date. Mixing them lets marketing wording read as verified performance.
When is the research finished?
When every row carries a capture date, scope, limitation, owner and recheck trigger, or is marked unresolved. Stop sooner if identity is unclear or collection would be disproportionate.
