Independent Ratings
All three sets of customer ratings for Arvell Watches - two hosted by a third party, one hosted by the brand - with score, review count, collection method and the date each was read.
There are three sets of customer ratings, and they are not the same kind of evidence. Two are hosted by a third party the brand cannot edit. One runs on the brand's own shop, is collected by inviting buyers after purchase, and is displayed by the brand.
All three are here, in one table, with the difference as a column rather than a footnote. A rating shown without its hosting and its date invites a comparison that cannot be made.
Demo figures - Arvell is a fictional brand. The structure, the columns and the rules below are the template; a real deployment replaces the numbers and keeps everything else.
The three sets
| Set | Hosted by | Collection | Rating | Reviews | Retrieved |
|---|---|---|---|---|---|
| Trustpilot | Third party | Self-selected | 4.4 / 5 | 212 | 2026-09-03 |
| Third party | Self-selected | 4.6 / 5 | 388 | 2026-09-03 | |
| On-site reviews | The brand's own shop | Solicited | 4.9 / 5 | 1,840 | 2026-09-03 |
Demo figures for a fictional brand; the structure is the deliverable. The two third-party sets answer 'how satisfied are customers'. The merchant-hosted set answers 'what do owners of this specific reference say', which the third-party sets cannot, because a Trustpilot review is about the company and an on-site review is attached to a named reference. The two kinds are never compared to each other on this site: a solicited score and a self-selected score are not measuring the same thing.
Why the highest score is not the headline
The on-site set is the largest and carries the highest score. That combination is why it is reported rather than led with. It is solicited, which means the sample is drawn from all buyers rather than from the people motivated enough to write unprompted, and it is displayed by the brand rather than by a party the brand cannot edit.
A higher solicited score is a well documented sampling effect. On its own it is not evidence of manipulation, and it does not rule it out either. What can be checked is checked below.
What the on-site set is genuinely better at is granularity. A Trustpilot review is about the company; an on-site review is attached to a named reference and often carries a photograph of that watch on a wrist. For "what do owners of this reference say", it is the strongest of the three. For "how satisfied are customers", the two third-party sets are the ones to cite.
The integrity test
A brand publishing its own review score has an obvious incentive problem, and asserting good faith does not address it. These five questions can be asked of any brand, and they are answered here for this one.
Five questions that can be asked of any brand's merchant-hosted review set, including this one. They test the published data for the patterns manipulation would leave, rather than asking the reader to take the number on trust. A brand that will not answer them has given an answer.
Why it catches something: A distribution that does not sum has had rows removed.
Yes. The demo distribution sums exactly to the stated total.
Why it catches something: Their absence in a set of this size is not plausible.
Yes, both are present.
Why it catches something: A solicited set scores higher, but a set with an order of magnitude fewer complaints has been filtered.
Yes, and the difference is stated as a sampling effect rather than as a quality finding.
Why it catches something: A number no one can open is an assertion.
Yes, the set is published on the shop with each review readable.
Why it catches something: A review count far above plausible order volume indicates imported or purchased reviews.
Unknown for the demo brand. Named here as unknown rather than answered.
What is still unknown
It is the difference between 'reviews carrying a verification label' and 'verified purchases'. The stronger claim is not available without the breakdown.
What would settle it: A per-status export from the review platform account.
Method
Scores are read directly from the platform, never aggregated or rounded up. Review counts are stated because a 4.8 from 12 reviews is not a 4.8 from 1,200. Checks run monthly and the retrieval date is part of the data rather than a detail below it. Which set answers which question is set out in what each source can and cannot show.
What customers praise and criticise is summarised, including the parts that are not flattering, in Strengths and Limitations.
Sources: Trustpilot profile (demo) · Google Business profile (demo) · On-site reviews on the brand shop (demo)
Last verified: 2026-09-03 · Last updated: 2026-09-03 · Maintained by: Arvell Watches - Brand Knowledge Team
GEO infrastructure by Concierca - machine-readable brand knowledge, kept current for AI systems.