Paper 1 · v3·May 8, 2026·6,338 words
Live · Observational·Last updated 20 Aug 2026·Latest: API vs. Search methods pilot
The evidence behind AI visibility.
An observational program measuring which businesses AI recommends and the signals that appear alongside visibility. 2 studies published, 2,729 businesses analysed (including the 1,004-business Phase 1 sub-sample).
Research reference library
V6, Wave 2 is planned for 17–21 November 2026. Results are pending. Read the launch preview or explore the methods and evidence already published.
2
Studies published
1 more in the pipeline
2,729
Businesses analysed
includes the 1,004-business Phase 1 sub-sample
3
Systems with complete coverage
ChatGPT, Claude, Perplexity (Gemini and Google AI Overviews for a subset)
14 · 4
Verticals · metropolitan markets
Los Angeles, New York, Chicago, Sydney
We're running an ongoing public research program on AI visibility across industries and markets.
AI search is rewriting how businesses get found. ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews now mediate billions of buying-intent queries per month — and the rules for being recommended are completely different from classical SEO.
Most published GEO advice is built on case studies of N=1, hot takes from a single account, or rebranded SEO playbooks. We wanted to know which signals appear alongside AI visibility — across thousands of businesses, multiple verticals and markets, with the methods, limitations and conflicts of interest disclosed.
So we built a research program. Same sample, transparent methodology, an open paper and dataset. The findings on this site are observational, and the full methods and limitations are available in the paper. Findings published whether they validate our product or not.
Key takeaways, methodology, and published papers are public. Individual brand lookup is free. We share aggregate findings and methods publicly while keeping the underlying business-level study dataset private. Contact the research team →
Six findings
Observational findings from the research program. Select a card to read the study.
Roughly seven in ten businesses (69.5% of 2,729) were not mentioned by any AI system they were tested on. Gemini and Google AI Overviews were tested for a subset of the sample only.
Read the study
Off-page presence (directories such as BBB, Yelp and Trustpilot, Wikipedia, LinkedIn company pages, and Reddit and Quora mentions) correlates with being mentioned by ChatGPT, Claude and Perplexity after nine controls.
Read the study
The single largest off-page partial r anywhere in the table is has_wikipedia → Claude (+0.219), and the second largest is quora_mention_count → Claude (+0.203).
Read the study
Businesses cited as a source by Perplexity were roughly six times as likely to be mentioned in that response.
Read the study
Domain Authority has a raw Pearson correlation of 0.337 with the composite visibility score on the Phase 1 sub-sample (n = 1,004) and 0.217 on the full sample (n = 2,200 businesses with non-null DA). After BH-FDR correction it remains significant as a raw correlation. In the multivariate L2-regularised logistic regression on n = 2,729 (which simultaneously estimates 22 features) its standardised coefficient is +0.139 with a bootstrap 95% confidence interval of [−0.031, +0.316], which crosses zero.
Read the study
Observational, not causal. The dataset cannot distinguish causation from confounding. Partial-correlation and OLS-residualisation analyses isolate linear contributions of a predictor controlling for a measured set of confounders; unobserved confounders (brand age, marketing spend, executive PR effort, demand-side prominence) are not controlled for.
Read the study
White papers
The findings of each study, repackaged in formal academic register — abstract, methodology, results, discussion, limitations, references. Cite these versions in research, journalism, and analyst work.
Phase 1 → 2 → 3
Each layer was designed to answer a question the previous one couldn't. The status of each is stated as it stands.
1
Published·April 2026 · superseded by the v3 paper (May 2026)
Study 1 — Outrigger AI Visibility Index
The Phase 1 baseline. Superseded by the v3 paper; the Phase 1 dataset record was withdrawn on 7 May 2026.
“This paper supersedes earlier reports of the same dataset. Version 1 (April 2026) reported headline statistics on a Phase 1 sub-sample of 1,004 businesses.” (v3 §1 Introduction). The v3 paper prints two Phase 1 figures: the sub-sample size (n = 1,004) and Domain Authority’s raw Pearson correlation of 0.337 with the composite visibility score on that sub-sample (v3 §4.4). No other Phase 1 figure is published here.
1,004 businesses in the Phase 1 sub-sample (v3 §3.2)
0.337 DA raw r on the Phase 1 sub-sample (v3 §4.4)
Withdrawn Phase 1 dataset record, 7 May 2026
Read the study
2
Published·v3 paper dated 7 May 2026
Study 2 — Outrigger AI Visibility Index: Off-Page Decomposition
A short summary of the published v3 paper. The paper on Zenodo is the authoritative source.
Off-page presence (directories such as BBB, Yelp and Trustpilot, Wikipedia, LinkedIn company pages, and Reddit and Quora mentions) correlates with being mentioned by ChatGPT, Claude and Perplexity after nine controls (v3 Tables 2–4).
2,729 businesses (v3 §3.2)
14 · 4 verticals · metropolitan markets (v3 §3.2)
3 systems with complete coverage: ChatGPT, Claude, Perplexity
Read the study
3
Did not launch·Pre-registered 7 May 2026
Study 3 — Layer 3: Controlled Intervention (Phase 3)
The controlled trial the v3 paper proposed to test causal magnitude. It was pre-registered, and it was never run.
Pre-registered on OSF (ZFA38, 7 May 2026); the trial did not launch; no businesses were enrolled.
ZFA38 OSF registration
0 businesses enrolled
What's new in the research
Every release, milestone, and operational note. Newest first.
- 20 Aug 2026Study published
Methods pilot published — ChatGPT API vs. ChatGPT Search
Across 500 sampled business-market comparisons, changing the collection route changed whether a business appeared visible in 16.4% of cases. Local services showed the largest gap at 27.5%.
Read the pilot - 8 May 2026Press
Academic preprint v3 published on Zenodo (DOI 10.5281/zenodo.20076380)
Cross-platform observational audit of 2,729 businesses across 14 verticals and four metropolitan markets. Supersedes v1 and v2; the v2 record (DOI 10.5281/zenodo.20048424) was withdrawn.
View Zenodo record - 7 May 2026Milestone
Layer 3 protocol pre-registered on OSF
Pre-registered on OSF (ZFA38, 7 May 2026); the trial did not launch; no businesses were enrolled.
View the OSF registration - 27 Apr 2026Study published
Study 2 web write-up published — Off-Page Decomposition
The original web write-up of the off-page decomposition. Replaced on 21 October 2026 by a short summary of the v3 paper.
Read the page - 10 Apr 2026Study published
Study 1 published — Outrigger AI Visibility Index baseline
Phase 1 baseline on a 1,004-business sub-sample. Superseded by the v3 paper; the Phase 1 dataset record (DOI 10.5281/zenodo.20048614) was withdrawn on 7 May 2026.
Read the study note
JH
Joel House
Founder of Outrigger (GEO platform). Founder of Joel House Search Media (one of Australia's largest SEO agencies by headcount). Forbes Agency Council. The research program is funded by Outrigger, with methods and conflicts disclosed — Joel publishes findings whether they validate our product or not.