Buyer Behavior Science
The Local-Intent Shift
Near-me searches no longer settle on one ranked page: they scatter across classic search, AI answers, and reviews, and the evidence shows that winning one surface is no guarantee you win the others.
Part of Choice Science in the Insights library.
Abstract
The direction of the shift is well established: people click through far less often once an AI summary appears in their results, and the share of US consumers turning to an AI tool for local recommendations jumped from 6% to 45% in a single year. What is not yet settled is the size of that AI-answer layer for local queries. Independent measurements of the same question range from 15% to 68% depending on definition, and no standardized, cross-checked benchmark exists yet for the industry to plan against.
Near-me discovery is fragmenting away from a single ranked page into three surfaces that behave, and rank, differently: classic organic and local-pack search, AI answers (Google AI Overviews, ChatGPT, Gemini, Perplexity), and the review and maps layer that supplies both. The scale of the move toward the AI layer is real, though its exact size depends on who is measuring it. Whitespark's audit of 540 local queries across three US metros found AI Overviews appearing for 68% of local-business queries overall, but that rate fell to roughly 15% for queries with strictly local intent, the very queries a nearby business cares about most. Semrush's larger tracking set (10 million-plus keywords through 2025) shows the same instability: AI Overview trigger rates swung from 6.49% in January to a 24.61% peak in July before settling near 15.69% by November, which argues against treating any single monthly snapshot as a stable baseline. The global picture, drawn from DataReportal and GWI's survey of 54 economies, shows the same current running underneath: the share of online adults who use a search engine to research a purchase fell from 51.3% to 45.8% over eighteen months, while consultation of an AI platform for that same research rose to 22.1%. Search still leads on pure brand discovery, 32.4% versus 14.8%, which is a useful reminder that the shift is uneven by task, not a wholesale replacement of one channel by another.
The behavioral change underneath these numbers is sharper than the adoption numbers alone suggest. Pew Research Center tracked 900 US adults across 68,879 real Google searches and found people clicked a traditional result only 8% of the time when an AI summary appeared, versus 15% when it did not, and clicked a link inside the summary itself just 1% of the time. People also closed out their search session more often after seeing a summary, 26% of the time versus 16%. For a local business, that is the crux of the shift: the moment where someone would have clicked through to your listing is increasingly a moment where they read an answer and never click anything. Reviews, meanwhile, are becoming more load-bearing, not less. BrightLocal's annual survey found 93% of consumers read reviews before visiting a local business and 74% now check two or more review sites before deciding, which raises the stakes on the FTC's 2024 rule banning fake and incentivized reviews, a rule with penalties of $51,744 per violation. Whether that same review layer is now filtered a second time through a shortlist an AI engine generates is a genuinely open question. SOCi's Local Visibility Index reported that ChatGPT actively recommended only about 1.2% of the 350,000-plus locations it tested and that 83% of restaurants tested never appeared in an AI recommendation at all, but the chain of custody on that figure runs through a commercial AEO-tool vendor's press release rather than SOCi's own report page, so we tier it as contested rather than treat it as settled fact.
The strategic implication is not that local businesses need to win a new channel, it's that they now need to win a second, differently-scored game with only partial overlap to the first. Whitespark's 2026 expert-consensus survey of local-SEO practitioners, the industry's longest-running study of this kind, weights Google Business Profile completeness at roughly 32% of classic local-pack ranking influence, review signals at about 20%, on-page SEO at 19%, and links at 15%, and for the first time added a separate factor category for AI-search visibility, an acknowledgment from practitioners themselves that this is now a distinct discipline. SOCi's contested but internally consistent overlap figure, roughly 45% overlap between businesses that rank well in traditional local search and businesses that get surfaced in AI recommendations, points the same direction: ranking well in the local pack does not reliably carry over. This is exactly the terrain the Visibility Corpus is built to map: not a single ranking, but where a local population's attention actually sits across search, AI answers, maps, and reviews, and how that attention is moving. A business that only optimizes for the local pack is optimizing for a shrinking share of the moment where the decision actually gets made.
The data, in one read
One page became three surfaces
For most of search's history, a local business had one job: rank on the page. That page is splitting into three surfaces that don't share a scoreboard. Classic organic and local-pack results still exist and still matter, Whitespark's own audit found the local pack appearing for 39% of local-business queries overall and 93% of queries with clear local intent. AI answers now sit alongside or above that pack for a meaningful share of the same searches. And underneath both, reviews and maps data feed the signals that decide who gets surfaced where. The starting point for any business owner is that these are not three names for the same thing measured three ways. They are three different competitions, and the evidence so far suggests a business can win one and lose the other two.
These are three different competitions, and a business can win one and lose the other two.
What people actually do when an AI answer shows up
The most rigorous piece of evidence in this study is also the simplest. Pew Research Center did not survey opinions about AI search, it watched what 900 real people actually did across 68,879 real Google searches collected over ten days in April 2025. When an AI summary appeared in the results, people clicked a traditional link 8% of the time. When it did not appear, they clicked 15% of the time, nearly double. Clicking a link cited inside the summary itself happened just 1% of the time. And people ended their entire search session more often after seeing a summary, 26% of sessions versus 16%. Put simply: the presence of an AI answer measurably reduces the chance that anyone clicks through to a business's own site or listing at all, for the exact query types where it appears. That is not a projection or a vendor's marketing claim. It's observed behavior from an independent research organization with no product to sell into the outcome.
A year of fast adoption, unevenly distributed
Adoption of AI tools for local decisions moved quickly in the US. BrightLocal's tracking survey found the share of consumers using an AI tool to find local business recommendations rose from 6% to 45% year over year, enough to put AI in third place among local-discovery channels behind only Google and Facebook; within that group, 31% had used ChatGPT and 23% had used Google AI Mode specifically for local recommendations. Similarweb's clickstream analysis over the following eighteen months shows the infrastructure catching up to that demand: ChatGPT's web visits grew 84% and its app usage roughly 156%, Gemini's web traffic grew close to ninefold with app usage up more than 1,100%, and total AI-platform referral traffic to third-party sites more than tripled in a year. Globally the pattern is real but more moderate. DataReportal and GWI's survey of 54 economies found the share of adults using a search engine to research a purchase slipping from 51.3% to 45.8%, while AI-platform use for the same task rose to 22.1%, and for pure brand discovery, search (32.4%) still comfortably outpaces AI platforms (14.8%). The shift is happening everywhere, but the US local-recommendation numbers are moving noticeably faster than the global purchase-research averages.
The paradox: AI answers are least common where local intent is strongest
This is the finding that should reshape how a local business owner reads every other statistic in this study. Whitespark's audit found AI Overviews appearing for 68% of local-business queries overall, a headline number that gets repeated everywhere. But when the query set was narrowed to strictly local-intent searches, the kind where someone is clearly ready to visit or call a specific type of business nearby, that appearance rate dropped to roughly 15%, while the classic local pack appeared for 93% of those same queries. Semrush's much larger tracking set (10 million-plus keywords) tells a compatible story from a different angle: AI Overview trigger rates were not steadily climbing through 2025, they swung from 6.49% in January to a 24.61% peak in July and back down to 15.69% by November. Two conclusions follow. First, the AI-answer layer is genuinely volatile month to month, not a fixed new floor. Second, for the queries that matter most to a local business, the classic local pack has not been displaced, it remains the dominant surface today. The strategic risk is not that AI has already taken over local search. It's that the AI layer is growing unevenly and unpredictably enough that no business can safely assume its current local-pack position is protecting it.
For the queries that matter most to a local business, the classic local pack has not been displaced. It remains the dominant surface today.
Two different ranking games
Whitespark's 2026 Local Search Ranking Factors survey, built on consensus from 47 practicing local-SEO experts across 187 factors, is the closest thing this space has to an established weighting: Google Business Profile completeness carries roughly 32% of local-pack ranking influence, review signals about 20%, on-page SEO 19%, link signals 15%, behavioral engagement 8%, and citations 7%. This was the first edition of the survey to add a standalone category for AI-search visibility impact, which the practitioners running it clearly now treat as a separate skill rather than a byproduct of ranking well elsewhere. SOCi's Local Visibility Index points at the same conclusion from the AI side, reporting only about 45% overlap between businesses that rank well in traditional local search and businesses that actually get surfaced in AI recommendations, with ChatGPT recommending just 1.2% of the 350,000-plus locations tested and 83% of restaurants never appearing in an AI recommendation at all. We tier that figure as contested rather than established, because the version of it we could verify traces through a commercial AEO-tool vendor's press release rather than SOCi's own report page, and the underlying methodology was not independently confirmed. The number may well hold up. Until someone traces it back to SOCi's own publication, it should be read as a strong signal, not a settled fact.
Reviews are becoming infrastructure, which raises the stakes
Reviews were already important before any of this; they are now doing double duty. BrightLocal's consumer survey found 93% of people read online reviews before visiting a local business, 82% read them regularly, and 74% now check two or more review sites before deciding, with Google itself the dominant review platform at 83% usage, ahead of local news sites, Yelp, and Facebook. Google's own ranking documentation confirms reviews feed directly into the "prominence" pillar of its three official local-ranking factors (relevance, distance, prominence), and Whitespark's practitioner survey puts review signals at roughly a fifth of local-pack ranking weight on top of that. Layer AI recommendation shortlists on top of an already review-dependent decision process, and the review layer becomes the closest thing local search has to a shared foundation across all three surfaces. That is exactly why the FTC's 2024 rule banning bought reviews, undisclosed insider reviews, fake suppression, and fake indicators of social-media influence, effective since October 21, 2024 with penalties of $51,744 per violation, is not a compliance footnote. It is protecting the one signal every surface in this study still relies on.
A parallel decline worth noting
One trend runs alongside this shift without clear evidence that it's caused by it. Pew Research's long-running tracking shows only 21% of Americans said they followed local news "very closely" in 2025, down from 37% in 2016, a steady decline that predates the current AI-answer surge by years. It would be tempting to draw a straight line from AI-mediated discovery to declining local-news attention, but the evidence available does not support that causal claim, the decline was already well underway before AI Overviews or ChatGPT existed as consumer products. A population paying less attention to local news is also a population whose local-intent decisions are more likely to be shaped entirely by whatever surface answers the question fastest, whether that's a map, a review score, or a synthesized answer.
The evidence, in numbers
Key findings, dated and sourced
-
Google AI Overviews appeared for an average of 68% of local-business queries across three US metros and six industries, but the rate dropped to roughly 15% for strictly local-intent queries (versus 92-97% for informational/hybrid-intent queries); the classic local pack appeared for 39% of the same query set overall and 93% on local-intent queries specifically.
emerging Whitespark case study, "The Prevalence of AI Overviews in Local Search" (540 queries across 3 cities x 6 industries), by Miriam Ellis (2025-05-12)
-
Tracking actual browsing behavior of 900 US adults across 68,879 real Google searches, users clicked a traditional search result only 8% of the time when an AI summary appeared, versus 15% when it did not, and clicked a link inside the summary itself just 1% of the time; users also ended their browsing session more often after a page with a summary (26%) than one without (16%).
established Pew Research Center, "Do people click on links in Google AI summaries?" (900 panelists, 68,879 searches, 12,593 with AI summaries, collected April 7-17, 2025) (2025-07-22)
-
Google states AI Overviews are live in over 200 countries and territories and 40-plus languages. The same post also claims an "over 10%" increase in Google usage for the query types where AI Overviews appear in the US and India; that growth figure is self-reported by Google about its own product with no disclosed methodology and should be treated as an interested claim, not an independent finding.
contested Google Search blog post on AI Overviews expansion (2025-05-20)
-
The share of US consumers using an AI tool (ChatGPT, Google AI Mode, etc.) to find local business recommendations rose from 6% a year earlier to 45%, making AI the third most-used local-discovery channel behind Google and Facebook; within that group, 31% had used ChatGPT and 23% had used Google AI Mode for local recommendations.
emerging BrightLocal, Local Consumer Review Survey, AI trust edition (n=1,002 US adults overall; AI-specific breakdown from a 455-respondent sub-panel) (2026-03-10)
-
93% of consumers read online reviews before visiting a local business and about 82% read them regularly; Google remains the dominant review platform at 83% usage, ahead of local news sites (48%), Yelp (44%), and Facebook (40%); 74% of consumers now check two or more review sites before deciding.
established BrightLocal, Local Consumer Review Survey 2025 (n=1,026 US adults) (2025-01-01)
-
An analysis of over 350,000 business locations across 2,751 brands found ChatGPT actively recommends only about 1.2% of local business locations tested, that 83% of restaurants tested never appeared in an AI recommendation at all, and that only roughly 45% overlap exists between brands that rank well in traditional local search and brands surfaced in AI recommendations. The chain of custody to the source organization's own report page was not independently confirmed; the figures were reproduced via a commercial AEO-tool vendor's press release.
contested SOCi, 2026 Local Visibility Index (350,000+ locations, 2,751 brands), as reproduced via National Law Review/MapAtlas syndication (2026-03-10)
-
A global survey spanning 54 major economies found the share of online adults who use search engines to research a purchase fell from 51.3% eighteen months earlier to 45.8%, while consultation of AI platforms for purchase research rose to 22.1%; for brand discovery specifically, search engines (32.4%) still outperform AI platforms (14.8%).
established DataReportal / GWI, Digital 2026 Mid-Year Global Update Report (240,000+ respondents, 54 economies) (2026-01-01)
-
Standalone generative-AI usage was measured at 2.42 billion active users globally, 29.2% of world population, up 141% year over year, with ChatGPT alone at approximately 1.15 billion monthly active users.
established DataReportal / GWI, Digital 2026 Mid-Year Global Update Report (2026-01-01)
-
The FTC's Trade Regulation Rule on the Use of Consumer Reviews and Testimonials, passed 5-0 and effective October 21, 2024, bans buying or selling fake reviews, undisclosed insider reviews, fake review suppression, and fake indicators of social-media influence, with penalties currently set at $51,744 per violation.
established Federal Trade Commission (FTC), Federal Trade Commission final rule announcement (2024-08-14)
-
Google Maps and local Search results are ranked primarily on three factors: relevance, distance, and prominence, with prominence driven substantially by review volume and quality plus off-site trust signals.
established Google Business Profile Help documentation (2024-01-01)
-
A practitioner-consensus survey of 47 local-SEO experts across 187 factors, run continuously since 2008, weighted Google Business Profile signals at roughly 32% of local-pack ranking influence, review signals at about 20%, on-page SEO at 19%, link signals at 15%, behavioral engagement at 8%, and citations at 7%, and added a new "AI Search visibility impact" factor category for the first time.
established Whitespark, 2026 Local Search Ranking Factors Report (2025-11-01)
-
Only 21% of Americans said they followed local news "very closely" in 2025, down from 37% in 2016, continuing a long-running decline in local-news attention.
established Pew Research Center, Local News Fact Sheet (2025-01-01)
-
Large-scale tracking of 10 million-plus keywords for AI Overview share found trigger rates were volatile rather than steadily climbing across 2025: 6.49% of tracked queries in January, spiking to 24.61% in July, then settling near 15.69% by November.
emerging Semrush AI Overviews Study (2025-12-15)
-
Clickstream analysis from September 2024 to March 2026 shows ChatGPT web visits grew 84% and app monthly active users grew roughly 156%, Gemini web traffic grew roughly ninefold with app monthly active users up over 1,100%, and total AI-platform referral traffic to third-party websites grew more than threefold between September 2024 and September 2025.
emerging Similarweb, Gen AI Stats 2026 (clickstream analysis) (2026-05-28)
Learning outcomes
What this study teaches
- Do not treat your local-pack ranking as proof you are visible in AI answers; the evidence shows these are two different competitions with only partial overlap in what earns a place in each.
- Reviews are the closest thing to shared infrastructure across every local-discovery surface right now, on-site conversion, local-pack prominence, and AI-answer trust all lean on them, which makes review authenticity a compliance issue as much as a marketing one under the FTC's current rule.
- Be skeptical of any single "share of AI answer" number you're handed, including some in this study; independent measurements of the same question currently disagree by a factor of four or more depending on definition and timing.
- The decline in click-through when an AI summary appears is real and measured, not a vendor scare story, which means the moment a customer used to spend on your website or listing is shrinking for a meaningful share of local queries, whether or not your rankings have changed at all.
- Local news attention has been declining for years independent of the AI-answer shift, so don't assume every change in local attention you observe is caused by AI; some of it is a much older trend running in parallel.
Honest limits
What this does not yet settle
- No independent or peer-reviewed study was found that directly measures the volume shift in near-me query behavior itself; every quantified figure on local-intent volume in this study is vendor- or platform-reported, each with its own methodology, and none are cross-validated against one another.
- Widely repeated historical local-search statistics (such as the often-cited "900% growth in near-me searches" figure) could not be traced to a live, dated primary source in this pass, despite being among the most frequently cited numbers in the industry, and were excluded rather than repeated secondhand.
- No standardized census of the AI-answer layer for local queries currently exists. The three studies that attempt to measure it (Whitespark, Semrush, SOCi) use different query sets, time windows, and definitions of "local," and their numbers cannot be reconciled into one trustworthy baseline.
- No data was located on messaging-app-based local commerce (for example, booking or inquiry activity through business messaging apps) or on dedicated maps-app usage as a discovery surface distinct from web search, a real gap given how much of local decision-making happens inside a maps interface.
- Commonly circulated local-pack click-through-rate figures could not be traced to a transparent, retrievable primary methodology and were deliberately excluded as unverifiable rather than presented as fact.
- A frequently repeated claim that a complete Google Business Profile drives several times more clicks or trust could not be traced to a live, dated Google source in this pass and was excluded rather than assumed true.
This is a synthesis of dated, attributed evidence, not a census. The AI-answer layer in particular has no independent, Nielsen-grade measurement yet, so readings of it are directional and named as a frontier, never presented as settled.
Straight answers
Frequently asked questions
Is the shift away from clicking local search results actually happening, or is this a vendor scare story?
It's measured, not marketing. Pew Research Center tracked 900 people across 68,879 real Google searches and found they clicked a traditional result only 8% of the time when an AI summary appeared, versus 15% when it did not, and clicked a link inside the summary itself just 1% of the time. People also ended their search session more often after seeing a summary, 26% of the time versus 16%. That's observed behavior from an independent research group with nothing to sell into the result, which is why the study treats it as established rather than emerging.
How big is the AI-answer layer for local searches, exactly?
There is no single trustworthy number yet, and the study says so plainly. Whitespark's audit of 540 local queries found AI Overviews appearing 68% of the time overall but only about 15% for strictly local-intent queries, while Semrush's tracking of over 10 million keywords showed the trigger rate swinging from 6.49% in January 2025 to a 24.61% peak in July before settling near 15.69% by November. The three vendors that attempt to measure this (Whitespark, Semrush, SOCi) use different query sets, time windows, and definitions of "local," so their numbers cannot be reconciled into one baseline.
If my business ranks well in Google's local pack, does that mean I'm also showing up in AI recommendations?
Not necessarily, and the evidence points the other way. Whitespark's 2026 practitioner survey added a separate ranking-factor category for AI-search visibility for the first time, treating it as a distinct skill from classic local-pack ranking. SOCi's Local Visibility Index reported roughly 45% overlap between businesses that rank well in traditional local search and businesses surfaced in AI recommendations, though the study tiers that figure as contested since it traces through a vendor's press release rather than SOCi's own report page. Either way, the study's read is that these are two different competitions with only partial overlap.
What should a local business actually do about reviews given all this?
Treat reviews as shared infrastructure, not just a marketing nice-to-have. BrightLocal found 93% of consumers read reviews before visiting a local business and 74% now check two or more review sites, and Google's own documentation confirms reviews feed the "prominence" pillar of local ranking directly. Because reviews now underpin the local pack, on-site trust, and any synthesized shortlist all at once, the FTC's rule banning fake or incentivized reviews, with penalties of $51,744 per violation effective since October 2024, makes review authenticity a compliance matter as much as a marketing one.
Is the decline in local news attention being caused by people turning to AI for local answers?
The study is explicit that this hasn't been shown. Pew Research's tracking shows only 21% of Americans followed local news "very closely" in 2025, down from 37% in 2016, a decline that was already well underway years before AI Overviews or ChatGPT existed as consumer products. The conclusion the study draws is not that AI caused the decline: a population already paying less attention to local news is more likely to have its local decisions shaped by whichever surface answers fastest, whether that's a map, a review score, or a synthesized answer.
Provenance
References
- Pew Research Center, "Do people click on links in Google AI summaries?", July 22, 2025 https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/
- Pew Research Center, Local News Fact Sheet, 2025 https://www.pewresearch.org/journalism/fact-sheet/local-news-fact-sheet/
- Whitespark, "The Prevalence of AI Overviews in Local Search", May 12, 2025 https://whitespark.ca/blog/case-study-the-prevalence-of-ai-overviews-in-local-search/
- Whitespark, 2026 Local Search Ranking Factors Report, November 2025 https://whitespark.ca/local-search-ranking-factors/
- BrightLocal, Local Consumer Review Survey (AI trust edition), March 2026 https://www.brightlocal.com/research/lcrs-ai-trust/
- BrightLocal, Local Consumer Review Survey 2025 https://www.brightlocal.com/research/local-consumer-review-survey-2025/
- SOCi, 2026 Local Visibility Index, as reproduced via National Law Review/MapAtlas, March 2026 https://natlawreview.com/press-releases/ai-search-recommends-only-12-local-businesses-rest-are-invisible
- DataReportal / GWI, Digital 2026 Mid-Year Global Update Report, 2026 https://datareportal.com/reports/digital-2026-mid-year-global-update-report
- Federal Trade Commission, final rule on consumer reviews and testimonials, August 14, 2024 https://www.ftc.gov/news-events/news/press-releases/2024/08/federal-trade-commission-announces-final-rule-banning-fake-reviews-testimonials
- Google, Business Profile Help, local ranking factors documentation https://support.google.com/business/answer/7091
- Google, Search blog, AI Overviews expansion update, May 20, 2025 https://blog.google/products-and-platforms/products/search/ai-overview-expansion-may-2025-update/
- Semrush, AI Overviews Study, December 15, 2025 https://www.semrush.com/blog/semrush-ai-overviews-study/
- Similarweb, Gen AI Stats 2026, May 28, 2026 https://www.similarweb.com/blog/marketing/geo/gen-ai-stats/
Every measured figure is dated to its capture and tagged with an evidence tier. Every cited work is real and locatable. Where an engine could not be captured this round, it is named as uncaptured, not estimated. Small-sample readings are labelled as directional.