Context behind every perspective.

Reviews have distinct origins.

ReviewList distinguishes reviews submitted directly by our members from imported reviews. External reviews are published only while an approved authorization covers storage, republication, and commercial use. Original source identifiers and review URLs remain in private provenance records. Permission to access an API alone does not establish permission to publish its content.

Our directory is currently building review coverage. A business with no eligible reviews remains unrated. Listing a business is not an endorsement, and we do not invent reviews or source scores.

Understand review coverage.

The catalog includes customer reviews collected from external platforms under confirmed permission. Averages describe only the reviews published here and may not represent every customer or match a platform’s official score. External reviewer names and personal contact details are omitted. Contact-number promotions are withheld for manual review regardless of rating.

The snapshot has a 30-day publication window and no automatic refresh. Source pages may contain later edits or removals. Employee reviews from workplace platforms are not included in customer ratings.

Ratings stay with their source.

We show the arithmetic mean of eligible, deduplicated reviews in our database separately for each source. These figures may differ from a source’s own published score because of coverage, freshness, eligibility, or its proprietary methodology. We do not present these figures as the source’s official score.

Ratings on a 1–5 scale stay on that scale. Other documented scales are mapped linearly to 1–5 using 1 + 4 × (rating − minimum) / (maximum − minimum). The original rating and scale remain stored with provenance. Distributions round normalized ratings to the nearest whole star.

We do not currently publish a combined company score. Software product reviews remain attached to the relevant product, separate from reviews of its parent company.

One experience should count once.

We match source review IDs and source URLs, check normalized content, and look for closely similar content from the same author and date. Potential duplicate relationships retain an audit trail. Duplicate copies are excluded from public counts and source averages. Automated matching can miss or incorrectly group reviews; administrators can inspect entity mappings and deduplication records.

Verification has specific meanings.

“Email verified” means the reviewer controls a verified email account. “Source verified” reflects the original platform’s verification flag. “Purchase verified” requires a moderator to inspect supplied evidence. None of these guarantees that every statement in a review is accurate. Domain ownership is verified using a DNS TXT record; it does not establish product quality or safety.

AI themes require underlying feedback.

Analysis requires at least ten eligible written reviews. Every recurring theme must reference at least two distinct review IDs and exact supporting excerpts. We validate citations before saving the result. Reviews are treated as untrusted data, and cannot issue instructions to the analysis system. Summaries describe reported experiences and are not independently verified allegations.

We cache analyses and refresh after material changes: normally at least five newly available reviews or ten percent of the previous sample. Removed or unauthorized underlying reviews invalidate affected analyses. AI can still make mistakes; read the customer feedback.

A website check is a technical snapshot.

Checks use public DNS, registry RDAP information where available, and an HTTPS connection with certificate validation. Direct website requests resolve and validate addresses first and pin the connection to a public address to prevent access to private networks.

A working HTTPS connection means transport encryption is available. It does not prove the business is legitimate. A new domain, private registration, or missing customer reviews is never treated as evidence of fraud. We do not automatically classify a business as a scam.

Labels distinguish limited information, a technical concern that warrants verification, no significant technical concerns detected, and specific verified adverse evidence. We do not currently have an external adverse-signal provider connected; no such signal is inferred from an unavailable provider.

Quality comes before indexing.

Business profiles remain accessible while they are enriched. Search indexing requires sufficient eligible review coverage, verified identity or useful evidence, fresh information, and original analysis. The current scoring model assigns up to 25 points for review volume, 20 for independent source coverage, 20 for identity verification, 20 for grounded analysis, 10 for technical checks, and 5 for freshness. The default threshold is 65, with at least five company reviews and current information.

Thin profiles, internal search results, and unsupported comparisons are noindex. Sitemaps exclude profiles that fail these rules. Imported averages are not marked as first-party review rich results.

Trust isn’t a paid feature.

Subscription plans never change public ratings, automated website observations, review visibility, or moderation decisions. Business corrections and policy disputes are available without payment. Review moderation follows the same rules for positive and negative feedback.