Test Your Domain

Scored the way AI systems actually read it

Enter your apex domain and the audit runs the same methodology behind Top10Lists.us against your site: a live multi-model panel asks five major AI engines to evaluate your domain — both from their blind prior and after fetching your live content.

The result is a per-engine matrix plus an overall row. The overall row carries the headline median (with outlier-drop) for GEO, SEO, and ΔNPS™; each engine row shows that engine's own GEO and λNPS™-derived ΔNPS. It runs in about a minute and returns raw receipts you can verify from outside the system.

Five engines, one median — so no model swings the verdict

Every headline figure is the median across the five models with outlier-drop: a model is dropped from the median when its spread is extreme (max − min greater than roughly ten points), so a single divergent engine never determines the headline number.

Retrieval-grounded answer engine — live web grounding gives a current-day read.

Frontier general-purpose model with a broad training corpus.

Search-grounded model weighted toward Google index signals.

Strict-rubric grader — deductions are common and intentional.

Fifth voice; completes coverage of the top conversational engines.

Eight structural checks, five quantitative metrics

The audit evaluates 13 signals — 8 binary structural readiness checks plus 5 quantitative metrics — and reports the median across the five model families with outlier-drop. The structural checks ask whether the bot-facing surface is present and parseable; the quantitative metrics measure how much of it AI retrieval can actually use.

One 0–100 GEO score, and the gap AI optimization moves

The GEO score is a 0–100 measure of how accurately and consistently AI engines describe and recommend your brand. It is the median across the five-engine panel, with outlier-drop, and it computes for every audited property.

ΔNPS = λNPS − μNPS per engine. μNPS™ is the engine-independent reading of what people say online; λNPS is a given engine's favorable/unfavorable prior. A positive ΔNPS means AI engines surface favorable material at or above what people say online; a gap is coverage the translation layer can close.

λNPS is what an engine says from memory with no site context; the receipt-first verdict is what it says after reading your fetched content. When the evidence lifts the verdict above the blind prior, the gap is closed by retrievable, citable material — not by reputation.