Methodology

A score you can't interrogate is marketing, not measurement. This page documents exactly how the Vizible Score is computed — the numbers below are imported from the same code that runs every scan, so they can't drift from reality.

Three layers, one score

AEO — AI Model Answers (API)

We ask each of your selected models your prompts directly via API and parse the answers for: mention (binary gate), sentiment, position in the answer, prominence (dedicated section vs passing), and framing. This measures what the models know.

GEO — Surface Truth (what users see)

We fetch the real Google AI Overview for each prompt via live Google results (SerpAPI) — the actual text and the actual cited source links Google shows users — plus a grounded Perplexity answer. Cited = your domain appears in the source list (100), mentioned in the text (50), absent (0). This measures what users actually see, which is not the same as what an API returns.

SEO — Organic Search

Live Google rankings for your prompts: your position, featured-snippet presence. AI answers are trained and grounded on the same web — organic visibility still feeds AI visibility.

Profile weights

How much each layer matters depends on how your buyers research. Pick a scoring profile per subject; the composite is a weighted average.

Profile	AEO	GEO	SEO	Trust bonus	Rationale
B2B	35%	25%	40%	+12	SEO still drives 60%+ of B2B organic traffic. AI chat is mid-funnel validation. GEO growing.
B2C	35%	35%	30%	+8	AI Overviews appear heavily on consumer purchase queries. AEO and GEO nearly equal.
Personal	50%	30%	20%	+10	People are researched primarily via AI chat. SEO matters for long-form content.
Local	20%	25%	55%	+8	Google Maps, local SEO, and AI Overviews for local queries. AI chat less used for local discovery.

Intent multipliers

Being visible on a buying-intent prompt ("best X", "X pricing") is worth more than on an awareness prompt. Each prompt is intent-classified and its contribution scaled:

buy

1.25×

B2B profile

compare

1.15×

B2B profile

research

1×

B2B profile

awareness

0.9×

B2B profile

Citations & sources

Every run extracts the domains cited in AI answers about your space and tags each with its origin: model (appeared in API answer text), google_aio (a real Google AI Overview source link), or perplexity (grounded answer citation). Surface evidence outranks model text. Aggregated across runs, this powers the citation-gap view: the sources to go win.

Cadence & limitations

•Subjects on the Drux plan are scanned daily; score history is retained per run, so trends are comparable run-over-run.
•LLM outputs are stochastic: single-run scores carry noise. Judge the trend line and the citation set, not one run.
•AI Overviews are localized and personalized; we measure a US/English, signed-out baseline.
•Sentiment and framing are model-parsed classifications, not human labels — accurate in aggregate, fallible per instance.
•Scores are a diagnostic, not a guarantee of traffic or revenue.

Questions about the methodology?

Talk to us →