Welcome to FactRelay

FactRelay is an auditable GEO operations and verified brand knowledge practice: we measure how AI assistants describe a brand, verify those descriptions against company-approved facts and cited sources, fix what can be proven wrong, and retest like-for-like.

This documentation is the public, load-bearing part of that promise. The standards we sell by are published here so you can hold us to them.

Start here

1. What is FactRelay? — the product in one page: what we do, what we refuse to promise, and the four product surfaces.
2. What is Generative Engine Optimization? — the category, minus the hype.
3. The 14-day GEO Baseline protocol — exactly what a first engagement runs, in order.

The standards

These four pages define how every FactRelay conclusion is produced. They apply to paid work, to our own self-audits, and to anything we publish.

• Measurement standard — surfaces, sampling, panels, and why single screenshots prove nothing.
• Evidence and source standard — what counts as evidence, validity attributes, and citation tracing.
• Building a verified brand fact registry — approved facts vs candidates, scope, versions, expiry.
• Correction, publishing, and retesting — the action side: fact-constrained drafts, publishing logs, like-for-like retests.

Legal and boundaries

• Data sources and compliance boundaries
• China and global data regions
• Privacy notice · Terms of use

What this documentation is not

It is not a content farm. English publications grow deliberately: each page exists because an engagement needs it or a standard requires it. Platform-specific observation guides (ChatGPT Search, Google AI Overviews, Perplexity, Gemini, Copilot, Claude) arrive in the weeks after launch, each built on primary sources with dates.