Buyers ask ChatGPT, Perplexity, and Gemini to recommend a vendor — and the engine names a short list before they ever hit a website. If you’re not in that list, you’re not in the deal.
It’s a sample, not a reading — so it ships with a confidence interval. A measured zero (competitors surfaced, you didn’t) is a confident low, never confused with “no signal.”
Each non-anchor pillar is scored only from evidence the scan actually found — or flagged “insufficient,” never guessed. Before any of it counts, a deterministic Tier-0 technical gate checks whether AI can even retrieve you (robots.txt, raw-HTML visibility, indexability) — a hard failure caps the whole score at 40, because authority you can’t retrieve is authority AI never sees.
These are measurement weights (where you stand today) — deliberately separate from the investment model on the GEO operating-system page (where to spend effort).
You get a score, the gaps holding you back, and the highest-leverage fixes. v1 measures one engine (Claude + live web search); ChatGPT, Perplexity, Gemini & Grok per-engine breakdown is on the roadmap.
Enter a company — we run its category's questions through a live engine, measure whether it surfaces, and score the full nine-pillar model behind a Tier-0 technical gate. Live research, ~30–60s.