Frequently asked questions

Your query is normalised into a canonical company name, then used to run a set of targeted web searches across review platforms, court and regulator databases, newsrooms and filing archives.

Promising pages are fetched and reduced to clean text. That text — never our own opinion — is passed to an analysis step that extracts ratings, recurring language, notable comments, complaints, company replies, legal matters and press items, each tied back to the URL it came from.

Reports are assembled on demand from public sources. Recent search results may be reused for 10 minutes and extracted pages for 30 minutes to reduce repeated requests. Source timestamps show when pages were read.

The company-report reader checks robots.txt and paces page requests. It does not bypass paywalls, logins or browser challenges. Sources whose pages cannot be read are labelled as discovery only and are not used for detailed claims.

Candidate pages must match every meaningful token of the company name in the page title or URL before their data is accepted. Near-miss profiles — a similarly named local business, a subsidiary, a defunct listing — are discarded instead of blended in.

Review and employer platforms for ratings and customer or staff sentiment; courts and dockets for litigation; regulators and agencies for enforcement, safety and labour records; investor filings such as 10-Ks and 8-Ks for what the company itself discloses; company newsrooms for official statements and replies; and reputable news coverage for context.

Yes. Where a primary document exists — a filing, a docket entry, a regulator record, an official statement — it takes precedence over commentary about it, and it is the link we cite.

No. Accreditation is a paid membership status, not a rating, so it is shown as a sourced status (Accredited, Not accredited, or Not found) with a citation, separately from any letter grade or star rating.

An empty section means nothing verifiable was found for it in that run, not that nothing exists. Smaller and private companies commonly have thin review, court and filing footprints.

No. Every numeric or letter score in a report is reproduced from the platform that published it, with its own scale, review count and source link shown alongside it. We do not adjust, weight or rescale another platform's rating.

The overall score is an AI-generated summary of the available reputation evidence, not a published platform rating or a fixed arithmetic average. It is shown as N/A when no verified company ratings are available. Individual platform ratings retain their source links.

Comments, complaints and descriptive words are classified as positive, mixed or negative from the language of the quoted source material itself. The filters on a report count those classifications so you can read only one side, and the counts always sum to the items retrieved.

Automated retrieval and analysis can misread or miss things, and platform ratings can be gamed or stale. Treat a report as a starting map: open the cited primary source for any claim that matters before acting on it.