The evidence layer
Reviews are everywhere.Trustworthy consumer knowledge is not.
Google, Yelp, Trustpilot, BBB, Reddit, marketplaces, influencers, and editorial publishers can each provide useful signals. RealReviews is designed to show how those signals fit together, what was verified, where incentives exist, and how confident a consumer should be.
Open five tabs. Give each one a job.
The map tab
Google or Yelp can confirm that a place is active, show recent photographs, and expose the local review stream. I still leave two blanks beside it: who was verified, and how the reviewer was invited.
The company-history tab
Trustpilot and ConsumerAffairs help with dates, response habits, and company-level patterns. A long feed is useful; it is not a substitute for knowing which collection program produced it.
The dispute tab
BBB supplies a different record: complaint, reply, accreditation context. The unresolved line is whether the promised remedy actually happened.
The messy-story tab
Reddit and specialist forums are where odd failures and missing vocabulary often surface. They are leads, not verified transactions, and one vivid thread may be wildly unrepresentative.
The exact-product tab
On Amazon or another marketplace, match the variant and seller before reading the stars. Purchase labels and review programs belong in the note too.
The demonstration tab
A creator may show what a product actually does; G2, Capterra, and other vertical sites may ask the better category questions. Vendor ties, supplied products, sponsorship, and commission stay attached to that evidence.
The company’s own wall
Testimonials show what the business selected for publication. That is evidence of a chosen story, not evidence of the full distribution of outcomes.
What goes in the comparison notebook
| Margin note | What can be copied into the record | What stays unresolved |
|---|---|---|
| Identity and transaction | The label a source provides, kept in its original context | A public account is not proof of a purchase |
| Date window | First and last dates, plus any obvious burst in collection | Older and newer reviews may describe different operations |
| Commercial tie | Sponsor, supplied product, affiliate link, accreditation, or business tool | Disclosure alone does not measure its effect |
| Agreement | The specific claim repeated across independent source types | Shared wording or recycled posts are not independent confirmation |
| Dispute result | Complaint, response, promised action, and documented follow-up | A reply is not the same thing as a completed remedy |
| Confidence | A separate judgment tied to verification and source coverage | Review volume by itself cannot fill an evidence gap |
A comparison assembled at the kitchen table
Say I am checking a local service. The map listing gives me the current address, photographs, and a recent run of reviews. A forum thread names a failure mode I had not considered. The complaint file shows that the owner answers, but one promised refund has no follow-up. Before I rely on the guarantee, I open the company’s terms rather than repeating a reviewer’s memory of them.
Nothing in that stack gets averaged into a magic score. I write down the matching dates and location, mark the unverified account, keep the missing refund outcome open, and record who could earn money from my decision. If two sources disagree, the disagreement survives into the conclusion.
The tabs do not vote. They leave a record.
Three reasons to stop short of a verdict
First: the rating belongs to a different seller, branch, product variant, or review period. Second: the loudest claim has no transaction trail and no independent match. Third: the sources contradict one another and the missing fact would change the advice.
In any of those cases, RealReviews should lower confidence and show the open question. Attribution stays attached. Commercial ties stay visible. The reader should be able to retrace the conclusion—and reject it—without treating this site as the only source that matters.