56 sourced statistics
Most AI visibility claims skip the denominator, so this sourcebook keeps it attached. 31 figures come from a read-only pull of Cheers production covering 13 home-services companies and 958,754 AI answers, corrected on September 28, 2026; the other 25 come from 7 public studies we opened and read.
Use these to set a baseline, not to promise an outcome. Product usage, consumer habits, and model answers all move faster than a quarterly refresh.
56
numbered statistics
31
Cheers panel figures
25
figures from 7 public sources
Cheers panel
Every figure in this section comes from the corrected production pull of September 28, 2026, covering the 28 days ending September 2, 2026. Each answer is counted once and each company once, and questions that name the company itself are left out. Nothing here is derived arithmetic; each figure is checked against the results file when the site builds.
13 home-services companies had AI prompt tracking running in the 28 days ending September 2, 2026.
The first edition reported 119 organizations, which were account records, not companies; the corrected count also leaves out a Cheers test account. Cheers-owned and market-baseline organizations are excluded from every customer panel cut.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
The home-services panel recorded 958,754 AI answers to non-branded prompts in that 28-day window, each counted once.
The first edition reported 2,671,846 checks because the aggregate table repeats each answer under a location, as a prompt total, and under every parent account. Branded prompts, which name the company itself, are left out.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
The mean AI appearance rate across home-services companies was 25.2 percent.
Each company is measured once across all of its non-branded prompts, locations, and engines, then averaged, so a large location network cannot stand in for the panel.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
The 25th-percentile home-services company appeared in 12.8 percent of its checks.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
The median home-services company appeared in 29.5 percent of its checks.
Corrected September 28, 2026. Each AI answer counts once, each company once, and questions that name the company itself are left out.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
The 75th-percentile home-services company appeared in 34.9 percent of its checks.
With 13 companies a 90th percentile would sit between the two highest companies, so it is not published.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Pooled across every check, home-services companies were named in 28.6 percent of AI answers.
A pooled rate lets companies with more checks weigh more. The company figures above give each company one vote.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
The mean citation rate across home-services companies was 19.1 percent.
Appearance means the answer named the business. Citation means the answer linked a source attributed to it.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
All 13 home-services companies were named at least once during the window.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
The median home-services company was cited as a source in 20.2 percent of its checks.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
The 13 home-services companies in the panel track 183 locations between them.
Real estate and beauty and personal care are tracked too, but each has fewer than ten companies, so neither is published or compared.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Google AI Mode produced the highest median company appearance rate of the four engines at 36.0 percent.
The engine Cheers labels GAIO is Google AI Mode. The first edition called it Google AI Overviews and AI Mode.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Google AI Mode also produced the highest mean company citation rate of the four engines at 31.1 percent.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
ChatGPT's median company appearance rate was 24.1 percent, second of the four engines.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Gemini's mean company citation rate was 10.8 percent, the lowest of the four engines.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Perplexity produced the lowest median company appearance rate of the four engines at 13.9 percent.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Pooled across every check, Perplexity cited the business in 32.7 percent of answers, the highest pooled citation rate of the four engines.
The pooled figure leans on the largest companies. By company mean, Perplexity's citation rate was second to Google AI Mode's.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
ChatGPT's mean company citation rate was 11.5 percent.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Among the 10 companies with ranked ChatGPT answers, the median company's average position when named was 2.65.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Questions that name the tracked company itself are left out; kept in, they would lift the mean company appearance rate to 27.9 percent.
46 branded prompts from 3 companies, which return the company almost every time. They measure whether an engine repeats a name the user typed, not whether it recommends the business.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Best or top list prompts without a place name produced a 28.4 percent pooled appearance rate across 12 companies.
Best near me and price prompts are not published: 9 and 5 companies track them, under the ten-company floor.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
Best or top list prompts without a place name made up 273 of the 593 distinct non-branded prompts tracked for home-services companies.
Source: Cheers production panel pull, corrected. Published 2026-09-02. Checked 2026-09-28. Methodology
At the median home-services company, each location collected 9.7 new reviews a month over the 90 days ending September 2, 2026.
Company level: reviews divided by the company's locations with at least one review and by three, then the median across 11 companies. Counts include reviews that reached Cheers after the first pull.
Source: Cheers production review pull, corrected. Published 2026-09-02. Checked 2026-09-23. Methodology
At the 75th-percentile home-services company, each location collected 26.4 new reviews a month over the same 90 days.
With 11 companies the 90th percentile is not published.
Source: Cheers production review pull, corrected. Published 2026-09-02. Checked 2026-09-23. Methodology
Home-services owners answered 82.6 percent of the 4,870 reviews their companies received in that 90-day window.
Source: Cheers production review pull, corrected. Published 2026-09-02. Checked 2026-09-23. Methodology
The mean home-services company rating across reviews received in the window was 4.58.
Real estate and beauty and personal care each come from a single company, so no rating comparison is published.
Source: Cheers production review pull, corrected. Published 2026-09-02. Checked 2026-09-23. Methodology
92.4 percent of the reviews home-services companies received in the window came from Google.
Source: Cheers production review pull, corrected. Published 2026-09-02. Checked 2026-09-23. Methodology
The Cheers Market Baseline recorded 356,019 citations across 100 metros and four engines in 28 days.
The baseline is a Cheers-owned organization built to sample what engines cite. It is not a customer and its own appearance rate is not published. It recorded 49,677 engine answers in the window.
Source: Cheers Market Baseline citation pull. Published 2026-09-02. Checked 2026-09-23. Methodology
ChatGPT sent 44.7 percent of its home-services citations to directories and review platforms.
Source: Cheers Market Baseline citation pull. Published 2026-09-02. Checked 2026-09-23. Methodology
Google AI Mode sent 11.4 percent of its home-services citations to Google's own pages.
Source: Cheers Market Baseline citation pull. Published 2026-09-02. Checked 2026-09-23. Methodology
Across all four engines, 68.1 percent of home-services citations went to business websites.
Source: Cheers Market Baseline citation pull. Published 2026-09-02. Checked 2026-09-23. Methodology
Adoption
Usage figures come from Google and OpenAI's published records.
Google reported more than 2.5 billion monthly active users for AI Overviews in June 2026.
Source: New controls for website owners in AI search. Published 2026-06-03. Checked 2026-09-02. Methodology
Google reported more than 1 billion monthly AI Mode users in June 2026.
Source: New controls for website owners in AI search. Published 2026-06-03. Checked 2026-09-02. Methodology
OpenAI reported more than 700 million weekly ChatGPT users by July 2025.
Source: How People Use ChatGPT. Published 2025-09-15. Checked 2026-09-02. Methodology
ChatGPT users sent more than 2.5 billion messages per day by July 2025, according to OpenAI.
Source: How People Use ChatGPT. Published 2025-09-15. Checked 2026-09-02. Methodology
About 70 percent of consumer ChatGPT messages were unrelated to work by June 2025.
Source: How People Use ChatGPT. Published 2025-09-15. Checked 2026-09-02. Methodology
Asking for information or advice accounted for 49 percent of consumer ChatGPT messages in OpenAI's study.
Source: How People Use ChatGPT. Published 2025-09-15. Checked 2026-09-02. Methodology
Writing tasks accounted for 42 percent of work-related ChatGPT messages in OpenAI's study.
Source: How People Use ChatGPT. Published 2025-09-15. Checked 2026-09-02. Methodology
Engines
These figures describe search-result behavior, source patterns, engine overlap, and results from academic testing.
Pew found an AI summary on 18 percent of the Google searches in its March 2025 browsing dataset.
Source: Google users are less likely to click when an AI summary appears. Published 2025-07-22. Checked 2026-09-02. Methodology
Google users clicked a traditional result in 8 percent of visits with an AI summary versus 15 percent without one in Pew's study.
Source: Google users are less likely to click when an AI summary appears. Published 2025-07-22. Checked 2026-09-02. Methodology
Users clicked a source link inside a Google AI summary in 1 percent of visits measured by Pew.
Source: Google users are less likely to click when an AI summary appears. Published 2025-07-22. Checked 2026-09-02. Methodology
Search sessions ended after 26 percent of pages with an AI summary versus 16 percent of pages without one in Pew's study.
Source: Google users are less likely to click when an AI summary appears. Published 2025-07-22. Checked 2026-09-02. Methodology
88 percent of the Google AI summaries studied by Pew cited at least three sources.
Source: Google users are less likely to click when an AI summary appears. Published 2025-07-22. Checked 2026-09-02. Methodology
The median Google AI summary in Pew's dataset was 67 words long.
Source: Google users are less likely to click when an AI summary appears. Published 2025-07-22. Checked 2026-09-02. Methodology
AI summaries appeared for 53 percent of Google queries with 10 words or more in Pew's dataset.
Source: Google users are less likely to click when an AI summary appears. Published 2025-07-22. Checked 2026-09-02. Methodology
AI summaries appeared for 8 percent of one-word or two-word Google queries in Pew's dataset.
Source: Google users are less likely to click when an AI summary appears. Published 2025-07-22. Checked 2026-09-02. Methodology
82.6 percent of cited real-estate agents in Local Falcon's study appeared on only one AI platform.
Source: The Realtor AI Visibility Index. Published 2026-08-18. Checked 2026-09-02. Methodology
The GEO research paper reported visibility gains of up to 40 percent from tested content changes across 10,000 queries.
Source: GEO: Generative Engine Optimization. Published 2023-11-16. Checked 2026-09-02. Methodology
Local behavior
These studies look at local business discovery and whether the same business appears across engines and cities.
The U.S. Small Business Administration counted 36 million small businesses, or 99 percent of U.S. businesses, in fiscal year 2025.
Source: SBA 2025 Annual Report. Published 2025. Checked 2026-09-02. Methodology
45 percent of U.S. consumers had used generative AI for local business recommendations in BrightLocal's 2026 survey.
Source: Local Consumer Review Survey 2026. Published 2026-02-11. Checked 2026-09-02. Methodology
Local Falcon ran 37,500 AI searches across 100 U.S. cities and five AI engines for its 2026 Realtor index.
Source: The Realtor AI Visibility Index. Published 2026-08-18. Checked 2026-09-02. Methodology
91.5 percent of the real-estate agents in Local Falcon's 2026 index were never cited by an AI engine.
Source: The Realtor AI Visibility Index. Published 2026-08-18. Checked 2026-09-02. Methodology
Reviews
Consumer survey figures show how reviews shape local decisions and how AI-generated summaries enter that process.
97 percent of U.S. consumers said they read online reviews for local businesses in BrightLocal's 2026 survey.
Source: Local Consumer Review Survey 2026. Published 2026-02-11. Checked 2026-09-02. Methodology
47 percent of U.S. consumers would not use a business with fewer than 20 reviews in BrightLocal's 2026 survey.
Source: Local Consumer Review Survey 2026. Published 2026-02-11. Checked 2026-09-02. Methodology
74 percent of U.S. consumers cared about reviews written within the previous three months in BrightLocal's 2026 survey.
Source: Local Consumer Review Survey 2026. Published 2026-02-11. Checked 2026-09-02. Methodology
82 percent of U.S. consumers read AI-generated review summaries in BrightLocal's 2026 survey.
Source: Local Consumer Review Survey 2026. Published 2026-02-11. Checked 2026-09-02. Methodology
Reading next
Changelog
September 2, 2026
Corrected September 28, 2026. The panel figures first published from the pull of September 3, 2026 at 04:05 UTC counted 119 account records as brands and repeated AI answers two to three times. The corrected count also leaves out branded prompts (questions that name the company itself), a Cheers test account, and every answer from before July 20, when an enrichment defect was still counting competitor names as the tracked business. Figures that no longer have ten companies behind them were withdrawn. The correction note lists what changed. First edition: 31 panel figures and 25 figures from 7 public sources.
