Stop collapsing visibility into one score: separate discoverability, citation, absorption and revenue
A single “AI visibility score” hides four different links in the chain: being retrieved, being cited, being absorbed into an answer, and producing business.
Key numbers
- The survey decomposes GEO into a multistage pipeline: search activation → crawling and indexing → retrieval → reranking and context allocation → citation → prominence → factual absorption → fidelity → user behaviour.
- It recommends representing visibility as a vector — discoverability, citation, absorption and economic outcomes measured separately rather than merged into one number.
- Reason one: already-retrieved content can causally change its citation or use, but no reviewed technique shows a stable long-term causal effect on organic discoverability — the two stages behave differently.
- Reason two: commercial audits show low source overlap and large run-to-run variability, so collapsing them into a single score turns noise into apparent trend.
Method and sample
arXiv survey “Optimizing Visibility in Generative Engines: A Critical Survey of GEO (2023-2026)” (2026-07-15): across 45 studies it proposes a multistage formal model and a visibility vector separating discoverability, citation, absorption and economic outcomes, with an evidence hierarchy and a reproducible measurement protocol.
What it means for brands
“Can AI see you” is not one number: retrieved, cited, written into the answer and generating enquiries are four different things. Merging them hides where the problem is and whether a change did anything.
How SourceSight handles it
SourceSight reports mention rate, position, source citations and coverage per keyword and engine, then maps fixes onto the right stage — discoverability via off-site and structured data, citation via pages and third-party sources, absorption via content format, revenue via your own conversion tracking.
Limits
The four-part vector is an analytical framework proposed by the survey rather than a standard definition; tooling still measures citation and absorption differently, so cross-tool comparisons need care.
References
Run it on your own project
What the same prompts return across 10 engines is only knowable by running them.
Related research
Same engine, different surface, different answer: test app, web and API separately
On the same platform and the same questions, app and web cite systematically different source sets — and the API shows something else again.
Methods · 2026-10-07How to design a retest you can trust: five moves so noise does not pass as progress
AI visibility is a distribution, not a point: fix the prompts, repeat the runs, add controls, paraphrase, validate by hand — or you cannot attribute the change.