AI search and small businesses: what the data says (2026)

AI assistants now operate at the scale of search engines: OpenAI says ChatGPT has 900 million weekly users, and Google says its AI Overviews reach more than 2.5 billion people a month. Consumers are starting to use them to choose local businesses: 45% of US consumers did so in 2026, up from 6% a year earlier. The best evidence also shows that AI answers reduce clicks to websites and that AI tools recommend far fewer businesses than a normal Google results page.

This report brings together the most credible published data on AI search as of October 2026. Every figure links to the source we opened. Where a number comes from a company that sells marketing software, we say so. Where the evidence is thin, we say that too.

Key takeaways: the five most important findings

1. AI is now a mainstream way to find local businesses, at least in the US. 45% of US consumers used ChatGPT or another AI tool for local business recommendations in the past year, up from 6% the year before, making AI the third most-used source after Google and Facebook (BrightLocal, 2026). In the UK, 36.7% of consumers surveyed said they had used AI for local search in the past month (Yext, 2026).
2. The platforms are huge. ChatGPT has 900 million weekly active users (TechCrunch reporting OpenAI's announcement, 2026). Google's AI Overviews have over 2.5 billion monthly active users and AI Mode has passed 1 billion (Google, 2026).
3. AI answers reduce clicks. Google users clicked a traditional result on 8% of visits when an AI summary appeared, against 15% when one did not (Pew Research Center, 2025). On keywords with an AI Overview, the top-ranking page's click-through rate was 58% lower than expected (Ahrefs, 2026).
4. AI sends real but small amounts of traffic. AI platforms sent just over 1.13 billion visits to the top 1,000 websites in June 2025, compared with about 191 billion from Google Search (Similarweb, 2025). Similarweb later reported that AI referrals stopped growing between mid-2025 and January 2026 (Similarweb, 2026).
5. AI recommends very few businesses. Across about 350,000 locations of 2,751 multi-location brands, ChatGPT recommended 1.2% of locations, Perplexity 7.4% and Gemini 11.0%. By comparison, brands appeared in Google's local 3-pack for their top keyword 35.9% of the time (SOCi, 2026).

How many people use AI assistants and AI search?

Hundreds of millions of people use AI assistants every week, and billions now see AI-generated answers in Google every month. The numbers below are what the companies themselves have stated. They use different measures (weekly, monthly, queries), so they cannot be added up or compared directly.

Table 1. AI assistant and AI search usage, as stated by the companies

A few details matter when you quote these:

• ChatGPT. The 900 million figure rose from the 800 million OpenAI reported in October 2025, according to TechCrunch (2026). OpenAI's own post was not accessible to us, so this figure is as reported by TechCrunch.
• Google. Sundar Pichai said at I/O 2026 that "AI Overviews now has over 2.5 billion monthly active users" and that AI Mode "in just a year" had "surpassed 1 billion monthly active users" (Google, 2026). In July, Google said it now sends "billions of clicks to websites every week through AI features in Search" (Google, 2026). In October, it said AI Mode queries are "on average three times longer than traditional searches" (Google, 2026).
• Microsoft. We found no official consumer-only Copilot user count. The 100 million figure combines work and consumer use and is more than a year old.
• Perplexity. The most recent figure we could trace to the company is from June 2025. Newer 2026 figures circulate on statistics blogs without a primary source, so we left them out.

What about the UK?

More than half of UK adults now say they use AI tools, and most UK searchers regularly see AI summaries in Google. UK data comes from Ofcom (the communications regulator), from the Reuters Institute at the University of Oxford, and from a smaller vendor survey.

Table 2. UK adoption data

Two cautions on the Ofcom figures. First, Ofcom's text says "around 30% of keyword search queries" show AI Overviews, but the footnote to the same sentence gives 39.9%, so treat the share as roughly 30–40%. Second, the 13% ChatGPT reach figure (up from 9% in 2024) comes from Ofcom's Figure 18, which measures reach among UK online adults in May of each year, not frequent use.

The Reuters Institute also found that across six countries (Argentina, Denmark, France, Japan, the UK and the US), weekly use of any generative AI system nearly doubled in a year, from 18% to 34%. Weekly use of AI for getting information rose from 11% to 24%. ChatGPT was the most-used system in every country, with an average of 22% using it weekly (Reuters Institute, 2025). Trust in AI search answers was lowest in the UK of the six countries.

What this means for a small business. Your customers are already using these tools, whether or not they mean to. Most UK searchers now see AI summaries at the top of Google without asking for them, and more than half of UK adults say they use AI tools. Treat "what does AI say about my business?" as a question to check regularly, just as you would check your Google reviews.

Do customers use AI to find and choose local businesses?

Yes, and the share grew sharply in one year, though most of the detailed data is from the US. The largest regular survey of local consumers is BrightLocal's Local Consumer Review Survey, which asks 1,002 US adults each year how they find and judge local businesses.

The 2026 edition, published on 11 February 2026, found (BrightLocal, 2026):

• 45% used ChatGPT or other AI tools for local recommendations, up from 6% the year before. AI became the third most popular source of business recommendations.
• Use of Google reviews fell from 83% to 71%. Local news sites fell from 48% to 29%. Apple Maps rose from 14% to 27%.
• 97% of consumers read reviews for local businesses, and 41% "always" do, up from 29%.

BrightLocal's follow-up report on AI, published on 10 March 2026, adds detail (BrightLocal, 2026):

• 31% had used ChatGPT for business recommendations in the past 12 months. 23% had used Google's AI Mode.
• Use is highest among 30–44-year-olds (64%) and lowest among over-60s (24%).
• 63% of AI users trust AI tools' business recommendations, and 10% distrust them.
• 42% of all consumers, and 64% of AI users, trust AI platforms as much as reviews.
• 97% of AI users sometimes double-check AI recommendations against real reviews, and 42% always check reviews on the original platforms.

In the UK, Yext's survey of 600 consumers adds (Yext, 2026):

• 74% of UK AI users rated their trust in AI local recommendations at 4 or 5 out of 5.
• 24% had tried a new local business because an AI recommended it.
• After an AI recommendation, 49% searched Google or Bing, 43% visited the business's website, and 37% clicked through to the AI's sources.
• ChatGPT was the most-used AI tool in the UK, followed by Google Gemini and Microsoft Copilot.

These surveys have limits. Both BrightLocal and Yext sell local marketing services. Both are online surveys, which tend to reach people who are more comfortable online. Yext's UK sample is small (600) and the fieldwork dates are not published. BrightLocal's methodology page does not give fieldwork dates either.

What this means for a small business. Customers who use AI still check. Most go on to look at reviews, search Google or visit your website before they buy. So AI does not replace your reviews and website; it adds a step before them. If an AI tool doesn't mention you, you may never get to that step.

Do AI answers reduce clicks to websites?

Yes. Every independent study we found shows fewer clicks on searches where an AI answer appears, though they measure it in different ways. Google disputes the scale of the effect but has not published data.

Table 3. Studies of AI answers and clicks

Other findings worth knowing:

• People leave Google more often after an AI summary. Pew found users ended their browsing session after 26% of pages with an AI summary, against 16% without one (Pew, 2025).
• Longer, question-style searches trigger AI summaries more often. Only 8% of one- or two-word searches produced one, compared with 53% of searches of ten words or more and 60% of searches starting with a question word (who, what, when, why) (Pew, 2025).
• Clicks fell even without AI Overviews. Ahrefs found click-through on comparable keywords without an AI Overview also halved over the same two years, from 7.6% to 3.9%. Seer found a 41% fall in organic click-through on queries without AI Overviews. So AI Overviews explain part of the decline, not all of it.
• Being cited helps. Seer found brands cited in an AI Overview got 35% more organic clicks and 91% more paid clicks than brands that weren't cited, in Q3 2025. Seer notes this may not be cause and effect: stronger brands may simply get cited more often (Seer Interactive, 2025).
• Local searches behave differently. In a test of 540 searches across three US cities and six trades (including plumbers and dentists), AI Overviews appeared on only 15% of local-intent searches (such as "personal injury lawyers in Phoenix"), where the map "local pack" appeared on 93%. But AI Overviews appeared on 92% of informational searches (such as "how much do personal injury lawyers charge in Phoenix?") and 97% of mixed searches (Whitespark, 2025).
• A UK view. Ofcom cites Similarweb data showing that US users' Google search clicks fell after they first used ChatGPT, to an index of 0.74 of their earlier level by month 4 (Ofcom, 2025). Ofcom's text describes this as a 26% fall "six months after" first use, while its chart shows months 0–4, so treat the timing as approximate.

What this means for a small business. "Near me" style searches still mostly show the map pack, so your Google Business Profile matters as much as ever. The bigger change is in the questions customers ask before they're ready to book: "how much does a boiler service cost?" or "do I need a solicitor for probate?" Those now mostly get an AI answer, and fewer people click through. If you write helpful pages to answer those questions, expect fewer visits from them and aim to be one of the sources the AI cites.

How much traffic do AI assistants send to websites?

AI assistants send a fast-growing but still small share of web traffic, and the growth stalled in late 2025. Most of the measurement comes from Similarweb, which estimates traffic from a panel of devices, and from Adobe, which measures visits to retailers that use its analytics software.

Table 4. AI referral traffic

Points to note:

• The base is small. Adobe itself says "the base of users remains modest" (Adobe, 2026). Large percentage growth from a small base can still mean modest absolute traffic. (The release covers the 2025 holiday season and was published in January 2026.)
• Visits to AI tools kept growing, but referrals did not. Similarweb describes a "permanent structural decoupling" between visits to AI platforms and the referrals they send (Similarweb, 2026). In other words, more people use AI, but they click out less per visit.
• ChatGPT changed how it links. In May 2026 ChatGPT replaced footnote-style citations with clickable brand names. Similarweb reports that the share of ChatGPT referrals landing on homepages rose from roughly 26–29% before the change to about 62–63% by late May 2026 (Similarweb, 2026).
• Citation rates vary by industry. Similarweb puts the citation rate at about 23% for Travel and Hospitality and about 20% for Automotive, but under 4% for Professional Services. No time period is given for these figures (Similarweb, 2026).
• ChatGPT's lead is shrinking. ChatGPT's share of worldwide visits to generative AI websites fell from about 76% in June 2025 to about 53% in May 2026, while Gemini's rose from under 9% to around 27–28% (Similarweb, 2026).

What this means for a small business. Don't expect AI to become your biggest source of website visits soon. For most small businesses, Google will still send far more. The value of AI visibility is less about clicks and more about being named: if ChatGPT names you, the customer may simply phone you or search your name. Ask new customers how they found you, and add "ChatGPT or another AI" as an option.

Which sources do AI engines cite?

AI engines lean heavily on a small set of big platforms: Reddit, Wikipedia, YouTube and, for local queries, Google Maps, business websites, Yelp, Facebook and niche directories. The exact mix varies a lot between engines and between studies.

Table 5. Published studies of AI citations

The only local-specific study of sources we could find with a published method was small. BrightLocal ran 20 searches across 10 industries (such as "best dentist in Denver") on Google AI Mode, Gemini, Perplexity and ChatGPT Search (BrightLocal, 2025). It found:

• Yelp was a source in 33% of searches. Gemini never cited it directly.
• All four engines used directories. For dentists, ChatGPT drew only on dental directories; for lawyers, ChatGPT and Perplexity relied heavily on Super Lawyers and FindLaw (US legal directories).
• Business websites made up the "vast majority" of sources, though no percentage was given for this study.
• Google AI Mode relied on Google Business Profile as a primary source.

SOCi's data also suggests that reviews work as a filter. SOCi describes ChatGPT as having "a high bar of 4.3 stars for inclusion in local recommendations". Businesses recommended by Perplexity averaged 4.2 stars and by Gemini 3.9 (SOCi, 2026).

On how Google's own AI features choose sources, Google has documented that a page only needs to be "indexed and eligible to be shown in Google Search with a snippet," and that "there's also no special schema.org structured data that you need to add" (Google Search Central, 2025). Google says AI Overviews and AI Mode use "query fan-out", "issuing multiple related searches across subtopics and data sources."

What this means for a small business. Every study we found says AI engines gather information about local businesses from the same few places: your Google Business Profile, your own website, the big review sites and the directories for your trade. If those say different things, or your contact details are out of date, AI tools may get them wrong or leave you out. In the UK, the directories that matter will be different from the US ones in these studies (see "What we'll measure next").

Will AI replace traditional search? What the forecasts say

Analysts predicted a sharp fall in search volume, but Google says its query growth is the fastest in five years. Both can be partly true, because they measure different things.

In February 2024, Gartner published this forecast: "By 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents." Gartner analyst Alan Antin said generative AI solutions "are becoming substitute answer engines, replacing user queries that previously may have been executed in traditional search engines" (Gartner, 2024). Gartner's site blocked our access, so we took the wording from DIGIT (2024), which reproduces the press release.

We did not find a published Gartner check on whether the forecast came true.

Google, meanwhile, reported in October 2026 that "our year-over-year query growth rate is the highest it's been in the last five years" and that AI investments are "driving overall Search query growth, including commercial queries" (Alphabet, 2026). In July it said AI Mode "is driving an incremental increase in Search queries overall" (Google, 2026). So Google's query count now includes AI-style searches; it is not a measure of "traditional" search in Gartner's sense, and the two figures cannot settle the question between them.

What this means for a small business. Don't plan for Google to disappear. Plan for a Google that answers more questions itself, alongside AI assistants that some of your customers use instead. The practical work is much the same for both: accurate listings, good reviews and clear information on your website.

What should a small business do this week?

Based on the evidence above, here is where to start:

1. Ask the AI tools yourself. Type the questions a customer would ask (for example, "best electrician in Leeds" or "emergency dentist in Bristol open Saturday") into ChatGPT, Gemini, Copilot and Perplexity, and turn on Google's AI Mode. These are illustrative examples, not measured search volumes. Note whether you appear, who does, and which sources are cited.
2. Fix your Google Business Profile. Google AI Mode relies on it (BrightLocal, 2025), and the map pack still appears on most local-intent searches (Whitespark, 2025). Check your hours, phone number, categories and services.
3. Make your details match everywhere. ChatGPT and Perplexity were only about 68% accurate on business contact details (SOCi, 2026). Check your listings on the main review sites and your trade's directories.
4. Keep asking for reviews. AI users still check reviews (BrightLocal, 2026), and recommended businesses tend to have high ratings (SOCi, 2026).
5. Make sure Google can index your website, with clear pages on what you do, where, and for how much. Google says no special markup is needed for its AI features (Google Search Central, 2025).
6. Ask new customers how they found you, and include "ChatGPT or another AI tool" as an answer.

Methodology and sources

How we chose sources

We searched for data published up to 31 October 2026 on six themes: AI usage scale, consumer use of AI for local businesses, effects on clicks, AI referral traffic, AI citation sources, and forecasts. We applied these rules:

• Primary sources first. We used the company's own statement, the original survey or the original study wherever we could open it. Where the primary page blocked access, we used a reputable news report of it and say so next to the figure.
• Most recent data first. We preferred 2025–2026 data and used older figures only where nothing newer had a traceable source.
• Method must be stated. We included a figure only if we could say who was measured, how, and when. Where a detail was missing (such as fieldwork dates), we say so.
• Vendor research is labelled. Much of the data in this field comes from companies that sell SEO, analytics or local marketing tools (Ahrefs, BrightLocal, Peec AI, Profound, Seer Interactive, Similarweb, SOCi, Whitespark, Yext). It can still be useful, but it has an obvious interest, and samples are often client bases or panels rather than random samples.
• Excluded: statistics roundup blogs, figures we could not trace to an original source, and figures whose original source contradicted them.

Sources, with dates

Gaps in the evidence

• Almost no UK data on local AI recommendations. The detailed consumer and recommendation studies are from the US. UK figures cover general AI use, not choosing a plumber or a dentist.
• No published study separates independent businesses from chains. SOCi's index covers multi-location brands only.
• Few studies cover the engines side by side with a published, repeatable method. That is the gap our own study is designed to fill.

What we'll measure next

SearchCrow will run a repeatable study of which businesses AI engines recommend for local questions across the UK, and publish the method, prompts and results. Everything in this section is a plan. No results exist yet; the (coming soon: searchcrow’s own study) placeholders show what we will report and how we will measure it.

Research questions

1. How often do AI engines recommend independent small businesses, compared with chains, franchises and directories or marketplaces?
2. Which sources do they cite for UK local questions, and how does this differ by engine, trade and city?
3. How consistent are the answers when the same question is asked again, and how accurate are the business details given?
4. How closely do AI recommendations match Google's local 3-pack for the same search?

Engines

Five engines, each queried through the most consumer-like access available, with web search turned on:

1. ChatGPT (with search)
2. Google AI Mode
3. Google Gemini app
4. Perplexity
5. Microsoft Copilot

Google AI Overviews will be captured separately from the normal Google results page for the same queries, alongside the local 3-pack, as a comparison baseline.

For each engine, we will record the access method (API or interface), the model or version shown, and the date and time of every run. Where we use an API, we will re-run a random 10% of prompts in the consumer interface to check the results match. (coming soon: searchcrow’s own study)

Cities (20)

Chosen to cover all four UK nations and a range of city sizes: London, Birmingham, Manchester, Leeds, Liverpool, Sheffield, Bristol, Newcastle upon Tyne, Nottingham, Leicester, Southampton, Brighton and Hove, Plymouth, Norwich, Exeter, Glasgow, Edinburgh, Aberdeen, Cardiff and Belfast.

Business categories and prompts

Ten categories common among UK small businesses: plumber, electrician, NHS dentist, solicitor (conveyancing), accountant, estate agent, hairdresser, vet, restaurant, and driving instructor.

Each category gets three prompt templates, one for each type of intent. These are illustrative customer questions, not measured search volumes:

1. Recommendation: "best plumber in [city]"
2. Urgent or specific need: "emergency plumber in [city] available today"; "dentist accepting NHS patients in [city]"
3. Trust or comparison: "which plumbers in [city] have the best reviews?"

The full list of 30 templates will be fixed and published before data collection starts.

Volume per wave: 10 categories × 3 prompts × 20 cities × 5 engines = 3,000 prompts. Each prompt is run 3 times (on different days within the same week) to measure consistency, giving 9,000 answers per wave. We will repeat the full wave every quarter.

Collection rules

• Fresh session for every prompt: logged out or a clean test account, no chat history, memory or personalisation.
• Requests sent from a UK IP address. The city is always named in the prompt, so results don't depend on the device's location.
• The full answer text, every cited URL and screenshots of a 10% sample are stored.
• No follow-up questions; the first answer only.

How we'll classify businesses and sources

Businesses. Every business named in an answer is extracted and matched to a real entity using its website, its Google Business Profile and, where available, Companies House. Each is classified as:

• Independent: a business with three or fewer UK locations that is not part of a franchise.
• Chain or franchise: four or more UK locations under one brand, or any franchise outlet.
• Directory or marketplace: a platform listing many businesses (for example lead-generation sites, review sites, trade directories and the NHS "Find a dentist" service).
• Unverifiable: the business cannot be found, or appears to be closed. These are counted as possible errors.

Sources. Every cited URL is grouped by domain and type: the business's own website; Google Maps or Business Profile; review platform; trade directory or marketplace; social media; forum (such as Reddit); news or blog; government, NHS or regulator; other.

Two people will classify an overlapping 10% sample independently. We will report how often they agree (Cohen's kappa), aiming for 0.8 or higher, and settle disagreements before final analysis.

Accuracy check. For a random sample of 500 recommended businesses, we will compare the phone number, address and opening hours in the AI answer with the business's Google Business Profile and website.

What we will report

• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)
• (coming soon: searchcrow’s own study)

All percentages will be reported with 95% confidence intervals (Wilson intervals for proportions; bootstrap resampling by city for comparisons between engines). We will publish the prompt list, the classification rules and the anonymised answer-level data with the results.

Limitations we already know about

AI answers change with model updates, so each wave is a snapshot. API results may differ from what users see in apps. Classifying "independent" depends on public information that is sometimes incomplete. Ten categories and 20 cities cannot represent every trade or town.

How SearchCrow fits in

SearchCrow runs checks like the ones above for a single business: it asks the major AI engines the questions your customers ask, shows whether you're recommended and which sources they used, and suggests what to fix. You can do the same thing by hand with the steps in "What should a small business do this week?"; SearchCrow just does it regularly and keeps a record.

FAQ

How many people use AI to find local businesses? In BrightLocal's 2026 survey of 1,002 US adults, 45% had used ChatGPT or another AI tool for local business recommendations in the past year, up from 6% the year before. In the UK, a Yext survey of 600 consumers published in June 2026 found 36.7% had used AI for local search in the past month. Both are online surveys run by companies that sell local marketing services.

Do Google AI Overviews reduce clicks to websites? The best independent evidence says yes. Pew Research Center tracked 900 US adults in March 2025 and found they clicked a normal search result on 8% of visits when an AI summary appeared, against 15% when none appeared. Ahrefs found the top-ranking page's click-through rate was 58% lower than expected on keywords with an AI Overview. Google says overall organic clicks are relatively stable but has not published figures.

How much traffic does ChatGPT send to websites? Still very little compared with Google. Similarweb estimated AI platforms sent just over 1.13 billion visits to the world's top 1,000 websites in June 2025, up 357% on a year earlier, while Google Search sent about 191 billion. ChatGPT accounted for more than 80% of those AI referrals. Similarweb later reported that AI referrals stopped growing between mid-2025 and January 2026.

Does ChatGPT recommend small independent businesses? Nobody has yet published a large study of independents. The closest is SOCi's 2026 Local Visibility Index, which tested 2,751 multi-location brands and found ChatGPT recommended just 1.2% of their locations, against 35.9% appearing in Google's local 3-pack. SearchCrow's planned study of 20 UK cities will measure how often independents appear compared with chains and directories.

Did Gartner say search volume would fall 25% by 2026? Yes, as a forecast. In February 2024 Gartner predicted that by 2026 traditional search engine volume would drop 25%, with search marketing losing market share to AI chatbots and other virtual agents. It was a prediction, not a measurement. In October 2026 Alphabet said Google's year-over-year query growth was the highest in five years, so the forecast and Google's own figures point in different directions.

Sources

• Pew Research Center, 2025: Google users are less likely to click on links when an AI summary appears in the results
• BrightLocal, 2026: Local Consumer Review Survey 2026
• BrightLocal, 2026: LCRS 2026, AI and trust
• BrightLocal, 2025: Which sources does AI search use for local businesses?
• TechCrunch, 2026: ChatGPT reaches 900M weekly active users
• Google, 2026: Sundar Pichai's I/O 2026 keynote
• Google, 2026: Q2 2026 earnings call remarks
• Google, 2026: Q3 2026 earnings call remarks
• Alphabet, 2026: Q3 2026 earnings release (SEC Form 8-K, Exhibit 99.1)
• Google, 2025: AI in Search is driving more queries and higher quality clicks
• Google Search Central, 2025: AI features and your website
• Microsoft, 2025: FY25 Q4 earnings
• TechCrunch via NetChoice, 2025: Perplexity received 780 million queries last month, CEO says
• Ofcom, 2025: Online Nation 2025
• Ofcom, 2025: From apps to AI search: how the UK goes online in 2025
• Ofcom, 2026: UK adults' media and online lives revealed
• Reuters Institute, 2025: Generative AI and News Report 2025
• Yext, 2026: How UK consumers navigate local search in the age of AI
• Ahrefs, 2026: Update: AI Overviews reduce clicks by 58%
• Ahrefs, 2026: 38% of AI Overview citations pull from the top 10
• Seer Interactive, 2025: AIO impact on Google CTR, September 2025 update
• Whitespark, 2025: The prevalence of AI Overviews in local search
• Similarweb, 2025: AI referral traffic winners by industry
• Similarweb, 2026: AI search trends: what changed in the past year
• Similarweb, 2026: AI search stats 2026
• Adobe, 2026: 2025 holiday shopping season press release
• Profound, 2025: AI platform citation patterns
• Peec AI, 2026: Most-cited domains in AI search
• SOCi, 2026: 2026 Local Visibility Index press release
• SOCi, 2026: 2026 Local Visibility Index webinar slides
• Gartner, 2024: Gartner predicts search engine volume will drop 25% by 2026 (wording checked via DIGIT, 2024)