Every number we have published, in one place. This is the complete statistical record of Steady Demand's research into AI search, Google Business Profile and Local Services Ads – 192 figures drawn from 24 studies, tests and client case studies published since December 2024.
The work spans large-scale AI citation analysis (1,487 local-service queries run through Gemini and ChatGPT, 14,472 citations classified), controlled GBP tests, LSA performance benchmarks, and results from client engagements. Every stat links back to the study that produced it, where you will find the methodology and sample size. Figures are recorded exactly as published – nothing rounded, nothing restated.
AI Citation Ledger — What AI Actually Cites
Published Aug 6, 2026
1,487 local-service queries run through Gemini and ChatGPT across 50 US metros and 10 verticals, with 14,472 citations classified. The largest look yet at where AI answers about local businesses actually come from.
Study scale
1,487queries completed (of 1,500 targeted)1,487 queries, 14,472 citations↳ The wider field: Profound classified 27 million citations across ChatGPT, Gemini and AI Overviews
14,472citations collected and classified1,487 queries, 14,472 citations50US metros × 10 service verticals1,487 queries, 14,472 citations3phrasings per metro/vertical combination1,487 queries, 14,472 citations3,611unique domains cited1,487 queries, 14,472 citations4,410distinct businesses named by Gemini1,487 queries, 14,472 citationsJul 27–Aug 20, 2026collection window1,487 queries, 14,472 citationsGemini API + ChatGPT comparison passesWhere citations point (Gemini)
59.9%citations to the business's own website1,487 queries, 14,472 citations14.4%citations to social / forum sources1,487 queries, 14,472 citations13.7%of all citations go to reddit.com1,487 queries, 14,472 citations10.4%citations to directories1,487 queries, 14,472 citations10.3%combined local-service directory share1,487 queries, 14,472 citations
Ratings & reviews of AI-recommended businesses
4.75★average rating of AI-recommended businesses1,487 queries, 14,472 citationsvs 4.84★ for the plain-search baseline90%of AI picks rated 4.5★ or higher1,487 queries, 14,472 citations97%of AI picks rated 4.0★ or higher1,487 queries, 14,472 citationsvs 99.6% of baseline businesses919average review count for AI picks1,487 queries, 14,472 citationsvs 846 for baseline; medians 295 vs 32235%of AI picks have 500+ reviews1,487 queries, 14,472 citations2.0%of businesses got recommended in more than one metro1,487 queries, 14,472 citations87 businesses total93%Google Places match rate for named businesses1,487 queries, 14,472 citations
- Main report: What AI Actually CitesAug 6, 2026 (upd. Aug 20)
- Methodology & study designAug 6, 2026
- Interactive dashboardAug 16, 2026
- Raw dataAug 16, 2026
Grounding Drift — AI Citations Change Every Time
Published Aug 6, 2026
Byte-identical queries repeated back-to-back, hours later and the next day. AI answers are not a ranking you can hold; the classic local pack is roughly ten times more stable.
Repeat-query consistency
46%best-case domain overlap on immediate repeatrepeat runs of the same 1,487 queriesJaccard 0.463↳ Others found the same instability: half of cited content is under 13 weeks old
0.056Jaccard overlap of the model's internal search stringsrepeat runs of the same 1,487 queries0.410domain overlap after ~3.5 hoursrepeat runs of the same 1,487 queries0.380domain overlap next dayrepeat runs of the same 1,487 queries0.265three-repeat agreement rate within one dayrepeat runs of the same 1,487 queries72repeat-test calls across 4 rounds over 2 daysrepeat runs of the same 1,487 queriesAI vs the classic local pack
7.9%Gemini returns the same top business on repeatrepeat runs of the same 1,487 queries90.2%Google local pack returns the same top listingrepeat runs of the same 1,487 queries0.177Gemini loose full-set overlaprepeat runs of the same 1,487 queriesvs 0.884 for the local pack5,000pairwise Gemini round comparisonsrepeat runs of the same 1,487 queries4,466 local-pack comparisons
Phrasing sensitivity
40%average source agreement across three phrasings of the same questionrepeat runs of the same 1,487 queries3% → 87%word-overlap range between phrasingsrepeat runs of the same 1,487 queriesworst: HVAC/Birmingham · best: Locksmith/Cleveland12.7%overlap of Gemini fan-out queries with Google's People Also Askrepeat runs of the same 1,487 queries80%of fan-out queries score below 0.2 lexical overlap with PAArepeat runs of the same 1,487 queries
External benchmarks cited
Gemini vs ChatGPT — Two Different Internets
Published Aug 6, 2026
The same 1,487 queries through both engines. They almost never agree on sources, and ChatGPT's source mix changed overnight on August 20, 2026.
Agreement between engines
8.3%average cited-domain overlap between Gemini and ChatGPTthe same 1,487 queries, both engines↳ Semrush found the same split – Gemini names brands 83.7% of the time, ChatGPT 20.7%
4.2%of queries where both name the same top businessthe same 1,487 queries, both engines0.037overlap across the full recommended-business liststhe same 1,487 queries, both engines0.5% – 14.9%domain overlap range by verticalthe same 1,487 queries, both engineslowest: personal injury · highest: auto repairChatGPT's source mix before → after Aug 20, 2026
41.7% → 0.0%reddit.com citation share for local queriesthe same 1,487 queries, both engines15.9% → 42.8%citations to business websitesthe same 1,487 queries, both engines34.6% → 40.8%citations to general directoriesthe same 1,487 queries, both engines3.0% → 11.6%citations to local-service directories (Angi, HomeAdvisor)the same 1,487 queries, both engines86.4%decline in ChatGPT's Reddit citations overallthe same 1,487 queries, both enginesPromptwatch tracking
Reddit reliance by vertical (July 2026, pre-change)
72.2%dentist queriesthe same 1,487 queries, both engines60.1%auto repair queriesthe same 1,487 queries, both engines12.3%personal injury lawyer queriesthe same 1,487 queries, both engines60%of July queries mentioned Reddit in the answer textthe same 1,487 queries, both engines899 of 1,500
Verification pass
334citations checked post-change — zero from Redditthe same 1,487 queries, both engines150 queries across 5 metros6.7%Reddit share in a same-day 40-query Gemini samplethe same 1,487 queries, both engines19 of 284 citations1,650total post-drop ChatGPT queries checkedthe same 1,487 queries, both engines
- Gemini vs ChatGPT local citations comparedAug 6, 2026
AI Overviews vs AI Mode — Opposite Citation Behavior
Published Aug 22, 2026
8,000 local-service searches across 50 metros and 10 home-services verticals, 86,645 citation references analyzed. The two Google surfaces reward almost opposite things.
Study scale
7,990local-service searches (≈4,000 per surface)8,000 searches, 86,645 citations↳ Ahrefs measured what that costs: 58% lower click-through at position one
86,645citation references analyzed8,000 searches, 86,645 citations~3,995usable queries per surface8,000 searches, 86,645 citations3,992 AI Overviews · 3,998 AI ModeHow often each surface appears
48.8%AI Overviews surface rate overall8,000 searches, 86,645 citationsas observed by third-party collection — a manual browser check found AI Overviews substantially more often100%AI Mode surface rate overall8,000 searches, 86,645 citationsnear-definitional — a dedicated AI Mode endpoint answers by design21.8%AI Overviews on local-intent queries8,000 searches, 86,645 citationsAI Mode: 100%. Collection-method rate; see study Finding 0175.8%AI Overviews on informational queries8,000 searches, 86,645 citationscollection-method rate; see study Finding 01
What each surface cites (local intent)
70.7%AI Overviews → business websites8,000 searches, 86,645 citations12.4%AI Overviews → social sources8,000 searches, 86,645 citations79.8%AI Mode → Google Maps8,000 searches, 86,645 citations12.8%AI Mode → business websites8,000 searches, 86,645 citations42.3% / 22.7%AI Mode informational: business sites / Google's own properties8,000 searches, 86,645 citations95.5%cited domain matches the GBP-listed website8,000 searches, 86,645 citations6,006 citations — 19% of all business-site citations88.0% / 43.4%AI Mode → Google Maps: highest vertical (dentist) / lowest (personal injury lawyer)8,000 searches, 86,645 citationsvaries far more by vertical than by metro71.2%–84.8%AI Mode → Google Maps: range across all 50 metros8,000 searches, 86,645 citationsevery metro above 70%
Traits of cited pages — AI Overviews vs AI Mode
79.3% / 74.0%schema markup present8,000 searches, 86,645 citationsbusiness-site citations only, one count per page94.8% / 95.1%viewport meta tag present — no difference between surfaces8,000 searches, 86,645 citationscorrected 2026-09-09; the earlier 87.2%/81.9% gap was a page-composition artifact72.8% / 69.1%page contains statistics8,000 searches, 86,645 citationsbusiness-site citations only, one count per page78.9% / 74.4%visible phone number8,000 searches, 86,645 citationsbusiness-site citations only, one count per page39.2% / 34.3%linked social profiles8,000 searches, 86,645 citations4.77★ / 4.75★average star rating of cited businesses8,000 searches, 86,645 citations412 / 450median review count8,000 searches, 86,645 citationsmeans were 1,129 / 1,266, skewed by national brands21.6 yrsmedian domain age — identical on both surfaces8,000 searches, 86,645 citations
Raw data released
86,645citation rows published as the companion dataset8,000 searches, 86,645 citations10,550 unique domains. Downloadable and filterable.3,997queries behind the dataset8,000 searches, 86,645 citationsAcross 50 metros and 10 verticals.11.70%of the residual category bucket is still misclassified8,000 searches, 86,645 citationsStated openly. The 73 highest-volume misclassifications were hand-corrected first.
What Actually Predicts a ChatGPT Ad
Published Aug 21, 2026
1,865 queries on the free tier, incognito, fresh session each time. Ad presence tracks intent and phrasing, not topic.
Intent is the driver
58.3%ad presence on local-intent queries1,865 queries, free tier, incognito1.44× the 40.5% general-commercial control23.8% vs 58.0%naturalistic vs templated phrasing1,865 queries, free tier, incognito≈2.4× difference54.3% vs 30.8%normal markets vs small towns1,865 queries, free tier, incognito
Ad presence by category
75.8%dentists1,865 queries, free tier, incognito70.8%hotels1,865 queries, free tier, incognito68.3%lawyers1,865 queries, free tier, incognito68.3%restaurants1,865 queries, free tier, incognito67.5%plumbers1,865 queries, free tier, incognito63.3%florists1,865 queries, free tier, incognito53.3%museums1,865 queries, free tier, incognito52.5%bookstores1,865 queries, free tier, incognito44.2%furniture stores1,865 queries, free tier, incognito18.3%auto repair1,865 queries, free tier, incognito
Ad quality
ChatGPT's Local Results Aren't Coming From Foursquare
Published Aug 21, 2026
2,880 prompts across 12 verticals and 12 US markets, plus a 1,728-prompt cross-check on a second API surface, to test the widely repeated claim that ChatGPT's local data is Foursquare's.
Foursquare's actual share
0.00%Foursquare citation share on the primary surface2,880 prompts + 1,728-prompt cross-check↳ Local Falcon found 83% of restaurants absent from ChatGPT entirely
0.06%Foursquare share on the independent cross-check surface2,880 prompts + 1,728-prompt cross-check7.5% vs 5.7%ChatGPT business resemblance to Google vs Foursquare2,880 prompts + 1,728-prompt cross-check87.5%of downstream repetitions of the original claim stripped all caveats2,880 prompts + 1,728-prompt cross-checkoriginal May 2025 study: 60–70%, five citiesWhat is actually in the grounding layer
95.83%Yelp present in business-card grounding2,880 prompts + 1,728-prompt cross-checkbut only 1.04% cited, 2.08% retrieved-not-cited28.90%TripAdvisor in business-card grounding2,880 prompts + 1,728-prompt cross-check13.55% cited · 20.70% retrieved16.88%OpenTable in business-card grounding2,880 prompts + 1,728-prompt cross-check1.70% cited · 3.58% retrieved4,608total prompts across both surfaces2,880 prompts + 1,728-prompt cross-check72-hour window, US-English
LSA Performance Benchmarks
Published Apr 18, 2026
What separates accounts generating leads from accounts that don't, from managed-account performance data.
Levers that move lead volume
+43%more leads when message leads are enabledmanaged-account performance dataand +19% impressions↳ Sterling Sky tracked LSA visibility rising from 11% to 31% of queries in a year
53.7average leads for accounts at 85%+ responsivenessmanaged-account performance data+77% lead volume, +52% impression rate141.8average leads for accounts hitting their budget capmanaged-account performance dataover 4× non-constrained accounts0leads for accounts with fewer than 4 photosmanaged-account performance dataCustom Services on a Google Business Profile
Published Dec 15, 2024
A single-profile test of adding custom services, tracked from March 2023 through December 2024.
Result
9keywords moved into positions 1–15one profile, Mar 2023 – Dec 2024one entered the top 3↳ Local Falcon put a number on the bar: 47 reviews is the median to rank
72 hourstime to first measurable movementone profile, Mar 2023 – Dec 202424% → 29%visibility score, March to Decemberone profile, Mar 2023 – Dec 2024+5 points, stable for over a year21 monthstracked duration of the effectone profile, Mar 2023 – Dec 2024Local Services Ads — Client Results
Managed and whitelabel LSA accounts, with the numbers each case study reports.
Cleaning company — 2 years
5,586total leads, Sept 2023 – Nov 2025single client account$174,109total spendsingle client account~$31blended cost per leadsingle client account$29.29 at launch → $29.85 at end349,182total ad impressionssingle client account99.5% / 54.3%top impression share / absolute top rank at peaksingle client account99.03% / 49.21% in Nov 2025190 → 218monthly leads, first month to lastsingle client account
National HVAC franchise — 4 years
12 → 68active LSA accountssingle client account0leads, calls or billable activity at handoffsingle client account$5K–$46Kmonthly spend per brandsingle client account3verticals: HVAC, plumbing, electricalsingle client account
Moving company — 9 months
#12 → #1rank placementsingle client account37 → 147monthly calls at peak seasonsingle client account300%+ call volume increase$38–$39cost per lead, stabilizedsingle client account97–98%top impression sharesingle client account272 → 290reviewssingle client account
Garage door client — 12 months
22 → 56–62monthly leads, baseline to peaksingle client account60+ at peak$40.97–$49.63cost per lead rangesingle client account2,921 → 6,340ad impressionssingle client account407 → 574reviews; rating 4.1 → 4.3single client account98–99%top impression sharesingle client accountholding ~#4 rank
- Cleaning company — launch to 5,586 leads in 2 yearsSep 2023 – Nov 2025
- National HVAC franchise — 12 to 68 LSA accounts4 years, ongoing
- Moving company in a saturated market9 months, ongoing
- Garage door client — escaping LSA verification limbo12 months, ongoing
GBP Suspensions, Spam & Recovery
Reinstatements, spam removal and ranking recovery work, with the reported numbers.
Spam removal — Dallas–Fort Worth
~500fake listings in the initial sweepsingle client account100% success rate within 30 days1,200+spam listings removed in under a yearsingle client account~100/month sustained~3×increase in calls from the client's listingsingle client accountheld top three in the map pack
Account-level suspension recovery
48 hoursto reinstate all but four locationssingle client account4 days total−50%revenue vs the previous week during suspensionsingle client account−70%revenue vs the same period the prior yearsingle client account
Repeat suspensions — garage door
17 daysfirst reinstatement (Jan 7–24, 2020)single client account$2,000/daylost revenue while suspendedsingle client account$34,000 total from the first suspension3suspensions in one yearsingle client accountlater reinstatements took 1 and 2 days4 of 16audit points out of compliancesingle client account
Ranking recovery & franchise fixes
#15 → top 4ranking restored after a review transfer went wrongsingle client account~1,800 reviews moved between locations72 hoursto visible recoverysingle client account17%call-in orders at a mis-chained franchise locationsingle client accountvs 40–55% at sibling locations
Local SEO boost — Barbara Oliver Jewelry
AI Citation Ledger — September 2026 Re-run
The identical design re-run six weeks later: 1,500 queries, 10,229 citations. Reddit fell out of both engines, the directories took the share, and one of the original findings reversed outright. July 2026 stays as the frozen baseline; these are the September figures, not a correction to it.
Study scale
1,500queries completed (of 1,500 targeted)1,500 queries, 10,229 citationsJuly wave: 1,48710,229citations collected and classified1,500 queries, 10,229 citationsJuly wave: 14,4726.8sources cited per query, on average1,500 queries, 10,229 citationsdown from 9.7 in July2,474unique domains cited1,500 queries, 10,229 citationsJuly wave: 3,6113,738distinct businesses named by Gemini1,500 queries, 10,229 citationsJuly wave: 4,410Sept 6–9, 2026collection window1,500 queries, 10,229 citationsGemini via Vertex AI + ChatGPT comparison passes
Where citations point (Gemini)
49.6%citations to the business's own website1,500 queries, 10,229 citationsJuly: 58.8%19.1%citations to local-service directories1,500 queries, 10,229 citationsJuly: 10.4%17.7%citations to general business directories1,500 queries, 10,229 citationsJuly: 10.5%3.9%citations to social / forum sources1,500 queries, 10,229 citationsJuly: 14.4%1.7%of all citations go to reddit.com1,500 queries, 10,229 citationsJuly: 13.7% — the single most-cited domain in the July study
Where citations point (ChatGPT)
46.0%citations to general business directories1,500 queries, 10,229 citationsJuly: 39.1%27.2%citations to local-service directories1,500 queries, 10,229 citationsJuly: 6.7%10.5%citations to the business's own website1,500 queries, 10,229 citationsJuly, restated after the Sept classifier re-audit: 2.9%0.0%of ChatGPT citations go to reddit.com1,500 queries, 10,229 citationsJuly: 41.7% — a complete removal, first measured Aug 20
Ratings & reviews of AI-recommended businesses
4.76★average rating of AI-recommended businesses vs 4.82★ for the plain-search baseline1,500 queries, 10,229 citationsJuly: 4.75★ vs 4.84★91%of AI picks rated 4.5★ or higher1,500 queries, 10,229 citationsJuly: 90%98%of AI picks rated 4.0★ or higher vs 99% of baseline businesses1,500 queries, 10,229 citationsJuly: 97% vs 100%860average review count for AI picks vs 850 for baseline; medians 307 vs 3111,500 queries, 10,229 citationsJuly, restated after a sampling fix: 906 vs 84636%of AI picks have 500+ reviews1,500 queries, 10,229 citationsJuly: 35%92%Google Places match rate for named businesses1,500 queries, 10,229 citationsJuly: 93%
Repeatability and drift
43.0%citation overlap when the identical question is asked twice the same day1,500 queries, 10,229 citations497 queries, 10 round pairs42.7%citation overlap a day later — statistically identical to the same-day figure1,500 queries, 10,229 citationsdifference +0.2 pts, 95% CI -0.9 to +1.2. July's next-day decay did not reproduce41.2%agreement when the same question is reworded three ways1,500 queries, 10,229 citationsJuly: 40.3% — rewording costs about as much as simply asking twice83.3%of the time Google's local pack returns the same top business on a repeat1,500 queries, 10,229 citationsJuly: 90.2% — the control moved too5.1%of the time Gemini names the same top business on a repeat1,500 queries, 10,229 citationsJuly: 7.9%
Gemini vs ChatGPT
8.3%of cited domains match when both engines get the identical question1,500 queries, 10,229 citationsJuly: 8.3% — unchanged to the decimal4.9%of the time the two engines name the same top business1,500 queries, 10,229 citationsJuly: 4.2%783queries where both engines named a recognisable business1,500 queries, 10,229 citationsJuly: 1,099. Gemini now returns no business name at all on 47.4% of queries, up from 25.8%
Concentration
Posts From GBP Do Not End Up in AI Results
Published Sep 27, 2026
A fabricated word was placed in a Google Business Profile post, then searched for across Google, AI Mode, ChatGPT, Claude, Perplexity and Maps. Nothing found it.
Retrieval across surfaces
0 of 8Google Search and AI Overviews returned the GBP post6 surfaces, 43 searchesA fabricated word was embedded in a GBP post, then searched for across six surfaces.0 of 8Google AI Mode searches located the post6 surfaces, 43 searchesA fabricated word was embedded in a GBP post, then searched for across six surfaces.0 of 8ChatGPT attempts retrieved the post content6 surfaces, 43 searchesA fabricated word was embedded in a GBP post, then searched for across six surfaces.0 of 7Claude searches found the post6 surfaces, 43 searchesA fabricated word was embedded in a GBP post, then searched for across six surfaces.0 of 8Perplexity attempts located the post6 surfaces, 43 searchesA fabricated word was embedded in a GBP post, then searched for across six surfaces.0 of 4Google Maps queries returned the post6 surfaces, 43 searchesA fabricated word was embedded in a GBP post, then searched for across six surfaces.
The control worked
- Posts From GBP do NOT End Up in AI ResultsSep 27, 2026
AI Overviews Fuse Into the Business Listing on Mobile
Published Sep 26, 2026
Mobile searches for 100 local businesses by name. Where an AI Overview appears, it is usually fused into the business listing rather than sitting above it.
Mobile vs desktop
26% vs 6%of businesses showed an AI Overview from the name alone, mobile vs desktop100 businesses, 4 queries each100 local businesses across ten categories. Mobile, New York, same day, four query variations each.80%of those AI Overview results were fused into the business listing100 businesses, 4 queries each100 local businesses across ten categories. Mobile, New York, same day, four query variations each.99%of the 100 businesses triggered an AI Overview on at least one of the four searches100 businesses, 4 queries each100 local businesses across ten categories. Mobile, New York, same day, four query variations each.
Who gets cited
74%ranked top 10 for their own name but still were not cited by AI Mode100 businesses, 4 queries each100 local businesses across ten categories. Mobile, New York, same day, four query variations each.59%of results put social and review sites above the business's own website100 businesses, 4 queries each100 local businesses across ten categories. Mobile, New York, same day, four query variations each.
Accuracy and stability
4businesses where Google showed the wrong business100 businesses, 4 queries each100 local businesses across ten categories. Mobile, New York, same day, four query variations each.1 of 20brands changed layout day to day on a repeat test100 businesses, 4 queries each100 local businesses across ten categories. Mobile, New York, same day, four query variations each.
Download all figures as CSV – free to cite with attribution to Steady Demand and a link to the source study.
For third-party research on the same questions, see the Local Search & AI Industry Research Index.
Data last updated October 13, 2026
Download all figures as CSV — free to cite with attribution to Steady Demand and a link to the source study.
