Sample data policy
What we publish about real brands, and what stays on our own fictional demo brand.
Saylent audits real AI assistants and records what they actually answered. Publishing that fact honestly - who an assistant recommends, and which page it built the answer from - is reporting, the same category norm every AI-visibility tool in this space follows for real brands. Publishing our own negative or unverified opinion about a company we don't own is a different thing, and we don't do it. The line stays the same everywhere on this site and in the product:
No verbatim negative claim, and no generated fix, about a named company we do not own - anywhere public.
Layer 1 - Factual
Real brands, real categories
Share of voice, who is recommended per question and how often, which sources and page types the engines cite, and bot-access facts of public sites. Quotes are used only when attributed to the engine and the date, and only when neutral or positive.
Layer 2 - Full report
A brand we own, or a brand with written consent
Sentiment, risk claims, rival comparisons, and generated fixes - the complete picture a report gives its own subject. Until a brand consents or we own one, this layer's public sample is Kestrel Uptime, a fictional company we audit.
Layer 3 - Aggregate
No brand named
Cross-site studies - bot-blocking rates, citation-source patterns - with no opinion about any named company.
On every public sample
- The date the assistants were asked, and the model names that answered.
- A disclaimer that a report records what the assistants answered on that date - it is not our assessment of the company.
- Attribution on every quote: which engine, which date.
- A visible way to request a correction or a takedown.
- Every other company's name and website is replaced with a fictional one. Only two kinds of names stay real: public platforms anyone can publish on (Wikipedia, Reddit, Hacker News, Stack Overflow, GitHub, YouTube, Medium) and the infrastructure vendors a fix tells you to open in your own dashboard (CDNs, web servers, hosting platforms), which are never the subject of a claim.
Request a correction
If something published here about your brand is wrong, or you'd like it removed, open an issue on the repository. We read every one.
Open a correction request on GitHub