https://localairegistry.com/research2026-11-10 https://localairegistry.com/research/how-often-ai-recommends-your-business https://i.ytimg.com/vi/g0Q8h4WKVw8/hqdefault.jpgHow Often AI Recommends Your Business: A Better Measure of Local VisibilityAsked the same query twice, Gemini’s sources overlapped by only 0.46 and a median of 29% of the businesses ChatGPT named appeared in both answers, so AI visibility is best measured as how often a business is named across many answers.https://www.youtube.com/embed/g0Q8h4WKVw8642026-11-122026-11-10 https://localairegistry.com/research/how-often-ai-recommends-your-business/images/ai-recommendations-change-when-same-query-is-repeated.pngAsk the same question twice and Gemini's sources overlap by only 0.46, versus 0.78 for Google SearchWhen the same query was run twice, the sources Google Search drew on overlapped by 0.78, Google AI Overviews by 0.66 and Gemini by 0.46 (Grossman et al., SIGIR 2026). In a separate preprint census of Bali venues, the local businesses named overlapped by 0.45 for Gemini down to 0.22 for Claude (Pitenin, 2026). Compare bars within a panel, not across panels. https://localairegistry.com/research/how-often-ai-recommends-your-business/images/four-conditions-that-change-which-business-ai-recommends.pngFour conditions change which business AI recommends: real local doctors named rose from 10.8% to 63.9% with web search onSystem and configuration, place, wording and time each changed AI recommendations in published research. Each mini-chart comes from a different study and metric, so read each one on its own, not across quadrants. Findings selected and grouped by Local AI Registry. https://localairegistry.com/research/how-often-ai-recommends-your-business/images/ai-visibility-confidence-interval-by-number-of-answers.pngA business named in 40% of AI answers has a 95% range of 17% to 69% after 10 answers and 31% to 50% after 100Illustrative calculation by Local AI Registry (95% Wilson score intervals, assuming independent answers under fixed conditions), not study data. The range narrows only as answers accumulate. https://localairegistry.com/research/local-answer-gap2026-11-07 https://localairegistry.com/research/local-answer-gap/images/seo-vs-aeo-explained-illustration.pngSEO gets you into the list. AEO gets you into the answerSearch engine optimization (SEO) earns a place in a list of links. Answer engine optimization (AEO) earns a place inside the AI-written answer that appears before any links. https://localairegistry.com/research/local-answer-gap/images/public-ai-search-datasets-local-query-census.pngOnly 2 of 39,676 public AI-search research queries ask AI to find a local providerAcross four public AI-search datasets, 239 of 39,676 unique queries (0.6%) concern a physical business, and only 2 ask an AI to find a local service provider. https://localairegistry.com/research/local-answer-gap/images/local-business-searches-by-day-easter-2026.pngLocal “is it open” searches spiked the day after Easter185 of 271 local searches in the dataset were collected on April 6, 2026, the day after Easter, which is why most local answers in this study are about holiday hours. https://localairegistry.com/research/local-answer-gap/images/google-ai-overview-activation-by-search-category.pngLocal business searches trigger Google AI Overviews more than any other topic76.4% of local-business searches showed a Google AI Overview, nearly double the next category (science, 39.9%) and over five times the 13.7% average. https://localairegistry.com/research/local-answer-gap/images/local-search-ai-overview-rate-vs-question-searches.pngLocal searches got an AI answer 76% of the time, even more often than other questionsControlling for phrasing, local searches were still 12 points more likely to get an AI Overview than other question-style searches (76.4% vs 64.1%). https://localairegistry.com/research/local-answer-gap/images/how-google-ai-builds-a-local-answer-illustration.pngHow Google’s AI builds an answer about a local businessFor local questions, Google’s AI assembled answers mostly from news coverage (67.5% of citations), rarely from the business itself (6.5%), and usually ended by telling people to check with their local store. https://localairegistry.com/research/local-answer-gap/images/who-google-ai-cites-for-local-business-questions-heatmap.pngWho Google’s AI cites for local business questions vs everything elseIn local answers, Gannett newspapers supplied 28.3% of citations and appeared in 90.3% of answers. Businesses’ own websites supplied 6.5% of citations. https://localairegistry.com/research/local-answer-gap/images/top-domains-cited-in-google-ai-local-answers.pngThe 18 websites Google’s AI cited most for “is it open” questionstoday.com was cited in 62% of local answers and usatoday.com in 47%. Gannett and TEGNA domains are highlighted; no store’s own site made the top 18. https://localairegistry.com/research/local-answer-gap/images/local-news-overrepresented-in-ai-local-answers.pngTEGNA TV stations are 18 times more likely to be cited in local answers than in other answersRelative to other topics, local answers lean on TEGNA TV stations 18 times more, AOL/Yahoo/MSN 4.1 times and Gannett 3.9 times. Social media is used less (0.7 times). https://localairegistry.com/research/local-answer-gap/images/ai-local-answer-source-2x2-share-vs-lift.pngLocal gatekeepers: which sources dominate Google AI’s local answersGannett newspapers and other news outlets are local gatekeepers: large shares of local citations and over-represented versus other topics. Business sites sit near the bottom left. https://localairegistry.com/research/local-answer-gap/images/ai-citation-sources-by-local-question-type-heatmap.pngThe vaguer the local question, the less the business gets citedBusinesses’ own sites made up 8.8 to 10.9% of citations when a brand was named, but only 1.0% for generic “what’s open” questions, which is how people discover new places. https://localairegistry.com/research/local-answer-gap/images/ranked-vs-used-by-ai-illustration.pngRanked first, still skipped: why SEO alone isn’t AEOIn 147 of 157 brand questions the store’s own site ranked on page one, but the AI cited it in only 70. Ranking and being used by AI are separate outcomes. https://localairegistry.com/research/local-answer-gap/images/brand-website-ranked-vs-cited-google-ai-overviews.pngRanked on page one, skipped by the AI: brand websites in local answersBrand websites appeared on page one in 147 of 157 brand questions but were cited in 70. Walmart ranked in 11 of 15 answers and was cited once. https://localairegistry.com/research/local-answer-gap/images/brand-rank-vs-ai-citation-2x2.pngRanked and used vs ranked and ignored: a 2x2 of brand websitesMost brands ranked on page one nearly every time. What split them was use: Publix, Trader Joe’s and Sam’s Club were cited every time; Walmart, Starbucks and Chipotle were ranked but ignored. https://localairegistry.com/research/local-answer-gap/images/brand-by-source-heatmap-google-ai-overviews-easter-hours.pngBrand by brand: where Google’s AI got each chain’s hoursWalmart, the most-asked brand, had its own site cited in 7% of its answers. Starbucks and Chipotle were never cited for their own hours; Publix, Trader Joe’s and Sam’s Club were cited every time. https://localairegistry.com/research/local-answer-gap/images/ai-citations-from-outside-google-page-one-by-source.pngThe AI goes past page one for news, not for business websitesOnly 4.1% of business-site citations came from outside Google’s first page, versus 48.8% for AOL/Yahoo/MSN and 39.8% for TEGNA stations. A business site has to rank to be used. https://localairegistry.com/research/local-answer-gap/images/ai-answers-telling-people-to-check-local-store-waffle.png173 of 207 local AI answers told people to check with their local store83.6% of local AI answers sent the reader back to the store (“check your local store,” “call ahead,” “hours may vary”), against 2.6% of other answers. https://localairegistry.com/research/local-answer-gap/images/how-ai-answers-local-questions-national-framing.pngAI answers local questions with national facts65.7% of local answers spoke about “all locations” or “most stores” (vs 2.3% of other answers), and 83.6% hedged (vs 2.6%). National facts are the AI’s default for local questions. https://localairegistry.com/research/local-answer-gap/images/ai-local-answers-with-address-or-phone-waffle.png1 of 207 local AI answers named a street address or phone numberOnly 1 of 207 local AI answers gave a street address or phone number. The AI answered about chains, not places. https://localairegistry.com/research/local-answer-gap/images/ai-store-locator-claims-unsupported-rate.pngWhen Google’s AI says “the store locator says,” half the time it can’t show it16 of 31 local claims that cited a store locator, app or official site (51.6%) were not supported by any page the AI cited, against 6.7% of other local claims. https://localairegistry.com/research/local-answer-gap/images/what-ai-can-read-on-a-web-page-illustration.pngWhat Google’s AI can actually read on a pageNews pages the AI cited had about 500 readable words. Store locator pages were mostly scripts, and the median cited social post had 8 readable words. https://localairegistry.com/research/local-answer-gap/images/cited-vs-readable-ai-sources-2x2.pngCited vs readable: what Google’s AI could actually readNews pages were both cited and readable. Social posts were cited in 48% of local answers but 78% were thin (median 8 words). 11% of businesses’ own pages were thin. https://localairegistry.com/research/local-answer-gap/images/readable-words-on-ai-cited-pages-by-source.pngSocial posts the AI cited had a median of 8 readable wordsThe median cited social post had 8 readable words. Store locator pages were also thin: Trader Joe’s store search had 97 words and Cracker Barrel’s Easter page 18. https://localairegistry.com/research/local-answer-gap/images/ai-sources-rotate-same-verdict-illustration.pngDifferent wording, different sources, same answerThree ways of asking about Target on Easter returned the same verdict but shared only about a third of their sources. No single publisher owns a local answer. https://localairegistry.com/research/local-answer-gap/images/ai-answer-source-overlap-by-wording.pngAsk the same thing in different words and two-thirds of the sources change“Is Target open on Easter” and “Target Easter hours,” asked the same day, shared about 35% of cited sources on average. The verdict stayed the same; the sources rotated. https://localairegistry.com/research/local-answer-gap/images/ai-overview-claim-accuracy-local-vs-other-heatmap.pngLocal AI answers were more faithful to their sources than average87.8% of 2,253 local claims were supported by the cited pages, versus 84.5% of 95,767 other claims. Faithful to sources is not the same as accurate for your location. https://localairegistry.com/research/local-answer-gap/images/sources-per-ai-overview-answer-distribution.pngLocal AI answers cite fewer sources: 7.2 per answer vs 8.1Local AI Overviews cited 7.23 sources on average versus 8.1 for other topics, clustered at 5 to 8 sources. https://localairegistry.com/research/local-answer-gap/images/single-source-of-truth-for-ai-local-aeo-illustration.pngOne confirmed record, pushed everywhere AI looksBecause AI rotates between sources, a business’s facts need to match on its website, Google Business Profile, directories and press. Local AI Registry starts from one owner-confirmed record. https://localairegistry.com/research/local-answer-gap/images/local-aeo-priority-2x2-control-vs-citation-share.pngLocal AEO priority map: what AI cites vs what you controlThe sources AI cites most for local answers (news) are the ones businesses control least, and the source they fully control (their site) is cited least. AEO closes that gap from both sides.