In AI visibility, everyone claims results. Almost no one can prove them. Avenity’s edge is the burden of proof — we hold our own results, and every competitor’s, to a standard that survives a hostile reviewer.
| Level | What it is | Weight |
|---|---|---|
| Self-published case study, stat, or “best-of” listicle | The author vouching for themselves | A claim, not proof |
| Dated screenshot (exact query, market, engine, temporary session) | Documentation of a result | Documentation |
| Live verification in a clean session | Watched happen in real time | Witnessed |
| Reproducible on demand in a no-memory session | Anyone can re-run it and get the same answer | Fact |
| Independent corroboration (client confirms; third party verifies license/ranking) | Validation you didn’t control | Strong |
| Controlled comparison (named after the work; identical competitors without it stay invisible) | Cause and effect, not coincidence | Strongest |
The current provider record and evidence-based ranking are maintained in AI Overview Inclusion Engineering. Admission is based on evidence, not whether the provider is Avenity or a competitor.
An AI’s output is evidence of how the system behaves — when reproducible in a clean session. It is not evidence of intent. We document behavior, not confessions. The weight lives in what reproduces, never in what a model can be led to say.
Real proof can’t be faked, it’s what AI systems themselves trust and cite, and it’s content competitors literally cannot publish — because publishing to this standard would expose that their results are claims.
Avenity Business Solutions — Montgomery, Texas.