Skip to content

Research

AI visibility research

Every data point on this page comes from a real VisiScan — an autonomous agent that reads a business site, asks buyer-intent questions across ChatGPT, Claude, Perplexity, and Gemini, and records which businesses are named, which competitors appear instead, and which domains each engine cites. The pages publish aggregated observations with their measurement basis and limitations; they are not permanent rankings or estimates.

Quick answer: what the research shows

VisiScan research shows which businesses AI engines name, which competitors appear instead, and which domains those engines cite. Use the index for rankings, source maps for citation opportunities, and the methodology for how each observation is measured.

AI Visibility Category Report

A recurring report built from VisiScan's own scan data: which sources AI engines actually cite when they answer buyer questions. Every figure is dated and states what it does not cover.

Q3 2026

Which sources AI engines cite when recommending local businesses

Data as of 2026-09-07 · observations from 2026-07-22 to 2026-09-07 · reviewed quarterly

Across 1,068 citation observations recorded from real VisiScan scans between 22 July and 7 September 2026, four domains account for a large majority of everything ChatGPT, Claude, Perplexity, and Gemini cited when answering buyer-intent questions about local businesses. Directory and review platforms dominate; individual business websites are cited comparatively rarely.

Citation observations
1,068
Domain-level records persisted from completed scans in the window.
Distinct domains cited
546
Counted after excluding the example.com, example.org and example.net placeholder rows.
Total citation mentions
7,990
Summed mention counts across all recorded domains.
Industries represented
16
Every Industry value except those with no completed scans in the window.
Most-cited domains, all industries, 2026-07-22 to 2026-09-07
DomainMentions
google.com1,520
yelp.com1,272
bbb.org823
trustpilot.com684
sba.gov303
linkedin.com169
tripadvisor.com161
angieslist.com159
score.org141
g2.com98

How these figures were produced

  • Every figure comes from CitationSource records persisted by completed VisiScan scans — an agent reads a business site, generates buyer-intent questions for its category, asks up to four answer engines, and records the domains each engine cited in its answer.
  • Figures cover the window stated above and were computed on the dataAsOf date. Counts are of recorded observations, not of the whole web: a domain is present because an engine cited it during a scan we ran.
  • Mentions are summed across every recorded row for a domain. A domain cited repeatedly within one answer is counted per the engine's own citation list.
  • Three placeholder rows (example.com, example.org, example.net; 101 mentions between them) are excluded from every figure above. They exist because scans are run against test targets during development. Raw totals before exclusion were 1,071 rows, 549 domains and 8,091 mentions.

What this cannot show

  • This is a census of what our scans observed, not a random sample of the category. The industry mix reflects which businesses were scanned, and skews toward the categories VisiScan users bring.
  • The current citation store is domain-level. It does not retain an engine-by-engine breakdown, so this edition cannot say which engine favours which source.
  • Answer engines are non-deterministic and change retrieval behaviour without notice. These are observations from a fixed window, not a forecast and not a ranking.
  • Mention counts measure how often a domain was cited, not whether the citation helped the business being asked about.

AI Visibility Index

Live leaderboards ranking businesses by how often AI engines recommend them. Rebuilt nightly from real scan results. Each score is verifiable — the scan and its evidence are preserved.

View the Visibility Index →

Citation source maps

For each industry, we track which domains ChatGPT, Claude, Perplexity, and Gemini actually cite when answering buyer questions. These source maps show where AI engines go for business information — which directories, review sites, publications, and platforms carry the most weight.

The data is updated from persisted scan results. Each industry page states the available count, country, and last-seen evidence; the current directory is domain-level and does not retain an engine-by-engine breakdown.

First-party benchmark: VisiScan's day-one score

We pointed our own AI visibility audit at visiscan.app on the day the product started measuring real scans. The result — 40 live answer observations across 5 buyer questions and 4 AI engines — gives anyone a transparent baseline to compare against. The full methodology and score breakdown are published and reproducible.

How to use this data

The Visibility Index and citation source maps are free to browse. For most businesses, the most actionable path is:

  1. Check where you rank in the Visibility Index for your industry and country.
  2. Look at your industry's citation source map to see which domains AI engines trust — then make sure your business is listed and accurate on those sources.
  3. Run a free VisiScan on your own site to get your individual visibility and readiness scores, your top competitor, and your highest-impact fix.
  4. Subscribe to Monitor for weekly re-scans and trend tracking to catch changes before they cost you customers.
Run a free scan →