Aigeo Documentation

Aigeo asks the AI assistants your customers actually use — ChatGPT, Gemini, Claude, Perplexity, Google AI Overviews, Google AI Mode, Grok, DeepSeek and Doubao — the questions your customers actually ask, then measures whether your brand comes up, where it lands in the answer, how it gets described, and who gets recommended instead of you.

These docs explain exactly how that works. Every formula, every denominator, and every case where a number is deliberately left out of an average is written down here, so the dashboard isn’t a black box.

Start here Quickstart From a new account to your first visibility score in about ten minutes. Start here Core concepts Target, prompt, probe, run, engine, region — the words the whole product is built from. Methodology How scores are calculated The end-to-end path from one AI answer to the Final Score on your panel.

The measurement in one page

Unit of measurement

1 probe

A probe is

1 prompt × 1 engine × 1 region

Score per probe

0–100

Headline number

Final Score

Everything on the dashboard rolls up from a single unit: one probe — one prompt, sent to one AI engine, in one region. Each probe is graded independently against a rubric chosen by what the prompt was trying to measure, and the panel averages those grades.

Final Score = 0.8 × average AI Score + 0.2 × SEO Score With no Lighthouse audit available, the SEO term is dropped entirely rather than counted as zero.

Scoring methodology

AI Score Position, sentiment, frequency and verdict — the 0–100 grade for a single answer. Final Score How AI visibility and the technical health of your own site are blended. Prompt intents Why "Dyson vs Dreame" is scored by a different rubric than "best hair dryer". Averages & denominators What each average excludes, and why a failed grade is never counted as a zero. Engines & regions Which engines carry a real geographic signal, and which can't be compared across markets. Competitors How rival brands are extracted, resolved to a parent brand, and turned into share of voice. Citations & sources The three source types AI engines cite, and which of them you can realistically influence.

Working the product

Set up a brand Targets, markets and brand context — the inputs every later score depends on. Prompts & running a probe Build the prompt library, pick engines and markets, launch a run. Reading the panel Overview, competitors, sources, probes and region comparison, tab by tab. Site Audit Whether AI crawlers can reach, read and understand your site at all. Discover Turn website analysis, search trends and Reddit threads into probe-ready prompts. Content Generate benchmark and authority articles against a specific prompt. Distribution Media releases and social placement, ordered and tracked inside the app. Reports & exports Client-ready HTML and PDF reports, plus raw CSV and Excel extracts.

Reference

Metric reference Every metric, its raw range, how it is displayed, and which probes it counts. Plans & usage How runs are counted against your allowance, including the region rule. Glossary Short definitions for every term used in the app and in these docs. FAQ Why a score moved, why a probe shows a dash, why two regions look identical.

Reading the dashboard

Every metric card in the app has a small i next to it. Hover it for the short version of the same methodology documented here — useful when you just need to remember whether Sentiment is scored out of 15 or out of 100.