Local SEO for AI search: the complete guide

Generative Engine Optimisation, or GEO, is the newest part of local search, and it is being built in public with far less settled evidence than conventional local SEO carries.

Most of what circulates about it is either recycled SEO advice with “AI” prepended, or a specific number — a Foursquare sourcing share, a schema-lift percentage — that does not survive being traced to where it actually came from.

This guide works from the evidence that does survive that trace: a cross-platform study of nearly 350,000 business locations, a controlled test of schema markup against real AI citations, a confirmed data-licensing deal between OpenAI and Yelp, and consumer research on how much people actually trust what these systems tell them. Where the evidence runs out, it says so.

AI recommendations are far more selective than Google’s local pack — and the three assistants aren’t equally selective with each other

Start with the finding this whole guide sits on top of. SOCi’s Local Visibility Index 2026 ran a standardised prompt — “Can you recommend businesses of X type in Y market?” — across six platforms, covering roughly 350,000 locations belonging to 2,751 multi-location brands.

ChatGPT recommended 1.2% of locations tested. Google’s local 3-pack featured 35.9% of the same locations for equivalent queries. That is a business appearing in Google’s local results roughly thirty times more often than it gets named by ChatGPT.

The full picture, per our AI local search statistics, is more useful than that headline ratio on its own, because the three assistants are not interchangeable:

Platform	Locations recommended	Business data accuracy	How it's grounded
ChatGPT	1.2%	68%	Yelp data-licensing partnership (from July 2026), plus Bing-indexed web search
Perplexity	7.4%	68%	Open-web retrieval, observably weighted toward Yelp and TripAdvisor
Gemini	11%	100%	Native Google Maps / Business Profile grounding
Google local 3-pack	35.9%	—	Google Business Profile, Google's own local index

Source: SOCi, Local Visibility Index 2026 — data on approximately 350,000 locations across 2,751 multi-location brands, predominantly US.

ChatGPT

1.2 %

Perplexity

7.4 %

Gemini

11 %

Google local 3-pack

35.9 %

Scale 0–40%. ~350,000 locations across 2,751 multi-location brands, predominantly US. Source: SOCi, Local Visibility Index 2026

Perplexity is roughly six times more permissive than ChatGPT; Gemini is roughly nine times more permissive than ChatGPT and the most generous of the three, though still a third the rate of the local pack itself.

Separately from how often each assistant recommends a business at all, there is a second finding worth holding onto: fewer than half of the businesses leading Google’s local rankings also turn up in AI-generated recommendations for the same query. Ranking first on Google is not a reliable predictor of an AI mention.

Whatever these systems are actually weighting, it diverges meaningfully from what wins a conventional local ranking. That is the reason a guide specific to AI search exists at all, rather than this being a subsection of ordinary local SEO advice.

One limit is worth stating plainly and carrying through the rest of this guide: SOCi’s study covers multi-location brands with existing digital infrastructure, not independent single-location businesses. Whether an independent local business faces the same roughly 30× selectivity gap is unmeasured, and it isn’t safe to assume the number transfers unchanged.

The accuracy gap decides which platform is actually worth prioritising

Selectivity is only half the problem. The other half is what happens once an assistant does mention a business — and here the three platforms split even more sharply than their recommendation rates suggest.

SOCi measured business-data accuracy — hours, address, phone number — against verified source-of-truth records for each assistant. Gemini scored 100%. ChatGPT and Perplexity both scored 68%.

The gap has an obvious mechanical explanation. Gemini is grounded directly in Google Maps and Business Profile data, the same index the local pack itself draws on. ChatGPT and Perplexity retrieve from a mix of licensed data and the open web, where a business’s hours or phone number can be stale, duplicated, or simply wrong on the source page the assistant happened to pull from.

Gemini

100 %

ChatGPT

68 %

Perplexity

68 %

Scale 0–100%. Hours, address and phone number checked against verified source-of-truth records. Source: SOCi, Local Visibility Index 2026

An assistant recommending a business off the wrong hours or a disconnected phone number arguably does more damage than not recommending it at all. A customer who shows up to a closed shop or calls a dead line does not file that under “the AI was slightly out of date,” they file it under the business itself.

That reframes the practical question. It is not just “does the AI mention me” but “does the AI mention me correctly,” and those are measurably different outcomes on ChatGPT and Perplexity specifically.

The practical implication follows directly. Gemini’s accuracy is not a lever a business pulls, it is a byproduct of using data Google already controls — which is precisely the same Google Business Profile accuracy that has mattered for the local pack for years.

There is no equivalent direct lever available for ChatGPT or Perplexity today, because a business cannot correct their retrieval the way it can correct a Business Profile listing. What it can do is keep the sources those two assistants actually draw from — its own website and its Yelp presence specifically — as accurate as the Business Profile it already maintains, on the reasoning that inaccurate information anywhere in that chain eventually surfaces somewhere an assistant reads from.

Schema markup is not the lever most AI-SEO advice claims it is

This is the least comfortable finding in this category, and the one most worth sitting with, because a large share of the content selling “AI search optimisation” services leads with exactly the tactic the evidence contradicts.

Ahrefs ran a controlled test on 1,885 pages that added JSON-LD schema markup, tracked for 30 days either side of the change and compared against 4,000 control pages, drawn from an initial six million URLs examined, across three AI platforms.

The result was a change in AI citation rate of −4.6% to +2.4% — a small, statistically significant decline for Google AI Overviews, and changes statistically indistinguishable from zero for Google AI Mode and ChatGPT. The study’s own authors put it bluntly: “we can’t tell whether the schema did a tiny bit of good or nothing at all.”

One boundary on that finding is worth keeping. It tested pages already receiving 100+ AI Overview citations, so it says nothing about whether schema helps an uncited page get its first citation. But for the population it measured, the tactic did not move the number in the direction it’s sold on.

Google’s own documentation backs this up directly rather than by omission. Search Central’s guidance on AI features in Search states, as of its most recent update: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.”

More specifically still: “There’s also no special schema.org structured data that you need to add.” That is the platform itself saying the mechanism a large share of AI-SEO advice sells does not exist as described.

The number frequently used to justify schema investment — a claimed 44% lift in AI citations from adding schema and FAQ markup, usually attributed to BrightEdge — does not check out. A real, publicly findable BrightEdge statistic does carry the figure 44%, but for something unrelated entirely: how much more often Google’s AI Overviews criticise brands compared with ChatGPT.

No BrightEdge study establishing a 44% schema-citation lift could be located. On the balance of what’s checkable, it looks like a misattribution that hardened into received wisdom through repetition — one sitting directly against both Google’s own documentation and a controlled study, rather than merely lacking support for it.

So what does correlate with AI citation? Whitespark’s 2026 Local Search Ranking Factors survey introduced a new scoring dimension specifically for AI search visibility, put to a 47-person expert panel:

Factor	Score
Presence on expert-curated "best of" lists	179
Dedicated page for each individual service	170
Prominence on key industry-relevant domains	167
Quality and authority of unstructured citations (mentions with no link)	160
Authority of third-party review sites	156

Source: Whitespark, Local Search Ranking Factors 2026

Three of the top five factors are variations on being mentioned favourably somewhere other than the business’s own website: a curated list, an industry-relevant domain, a review platform, not on-page technical work. That is a genuinely different priority list from schema and structured data, which is where most “how to rank in ChatGPT” content still starts. It’s not hard to see why.

Schema is a task an agency can bill for and a business owner can point to as finished in an afternoon. Earning a mention on a genuine industry “best of” list or from a local publication is slower, less billable and harder to guarantee — and the evidence above says it currently matters more.

This isn’t an argument that schema is worthless. It still supports the machine-readability that underpins other things this site recommends, covered in the local business schema guide, and the same is true of the technical checks that genuinely affect whether a page can be read at all. It’s that schema shouldn’t be where the AI-citation effort or budget concentrates, because the controlled evidence says that specific expectation isn’t being met.

Where each assistant’s data actually comes from, and what it means for Yelp specifically

Sourcing behaviour in this category has moved fast enough in 2026 that a widely repeated statistic became outdated mid-year. It’s worth separating what’s officially confirmed from what’s inferred by practitioners testing outputs, because most coverage blurs the two.

The confirmed part: OpenAI and Yelp entered a non-exclusive data-licensing partnership, reported first on 23 July 2026 and confirmed by Yelp’s own blog on 10 August 2026.

It feeds over 330 million Yelp reviews across more than 8 million business listings directly into ChatGPT — reviews, ratings, photos and business details, plus in-chat Reservations and Waitlist functionality for thousands of restaurants, in an initial US and Canada rollout. That is an official announcement from both parties, not an inferred pattern, and the wider Yelp usage data says how much reach it carries.

The inferred part: no official documentation from Perplexity about how it sources or weights local business data was identified in the sources reviewed for this guide.

What is observable, rather than confirmed, is that practitioners testing Perplexity’s local outputs consistently report a heavy lean toward Yelp and TripAdvisor listings, with the assistant’s explanations for a recommendation frequently quoting or paraphrasing review text directly. That’s a pattern read off outputs, not a disclosed sourcing policy, and it should be cited with that distinction intact.

Put together: two of the three major assistants — ChatGPT by confirmed partnership, Perplexity by consistent observed behaviour — now appear to lean heavily on Yelp specifically, rather than the open web in roughly equal measure. Gemini is the outlier, grounded in data Google already controls.

The practical implication is that a complete, accurate, actively managed Yelp profile has gone from “one directory among many” to a plausible single point of leverage across two of the three assistants a local business is likely to be evaluated by. That is a different way of ranking the work in the citation-building guide, where Yelp has always sat as one entry in a long list.

It’s worth confirming directly for a given business — checking what ChatGPT and Perplexity currently say about it specifically — rather than assumed universally, since none of this is guaranteed to hold the next time a licensing deal changes or a model update shifts retrieval behaviour again.

It’s also worth actively correcting a claim that still circulates despite being wrong for the current period. A widely repeated figure holds that Foursquare supplies 60–70% of ChatGPT’s local business results. That number traces to a small, single-country test (50 prompts across five Spanish cities, published May 2025) whose author’s own qualifiers were stripped out within nine days of publication as the figure spread as an unscoped, universal fact.

A large, current US test — 2,880 prompts across 12 metro areas and 12 categories, run in 2026 after the OpenAI–Yelp partnership — found 0.00% Foursquare citation share on ChatGPT’s primary answer surface, with a cross-check on a second surface finding 0.06%.

Whatever was true of a small Spanish sample in May 2025 is not a description of how ChatGPT sources local results now. Anyone still allocating Foursquare-specific effort on the strength of the 60–70% figure is optimising for a claim the current evidence directly contradicts.

Google’s AI Overviews trigger far less often for “near me” searches than for informational ones

Most coverage of AI Overviews and local business treats “does an Overview appear” as a single number, usually cited around 68%. The one study that actually broke local queries down by intent found a much larger gap hiding inside that average.

Whitespark tested 540 queries across three US cities and six industries: plumbers, personal injury lawyers, dentists, optometrists, medical clinics and real estate agents.

Narrowly local-intent queries, the “plumber near me” kind, triggered an AI Overview about 15% of the time. Informational queries in the same trades and cities, such as “how do I know if I have a slab leak,” triggered one about 92% of the time. Hybrid queries mixing both intents triggered Overviews 97% of the time, the highest of the three categories, while the local 3-pack appeared for 39% of the same query set — roughly the inverse relationship you’d expect.

Read plainly, that means most of the AI Overviews concern in local SEO is currently overweighted toward the query type least likely to actually cost a business its click. A transactional “near me” search still mostly returns a conventional page of local pack and organic results.

The search far more likely to get answered inside an AI box, before the searcher reaches a business’s own site, is the informational one sitting above the commercial pages — the how-to and explainer content most local SEO effort still deprioritises in favour of service pages and “near me” targeting. That is the keyword-to-content mismatch showing up on a new surface.

This doesn’t mean local businesses get named directly less often than expected inside the Overviews that do appear. A sampled subset of Houston plumber queries found 60% of citations pointed to third-party publishers against 40% citing an individual local business directly — a real if small illustration that direct citation happens, not just referral through a publisher.

But it does mean the practical fix isn’t “defend the near-me query,” which is largely still working the way it always has. It’s building out the informational layer of the site — the explainer and how-to content answering the questions that lead up to the transactional search — because that’s the layer this data says is actually being intercepted.

One flag worth stating plainly: this Whitespark study is from May 2025, over sixteen months before this guide’s own review date, in a category that has moved substantially since (the sourcing section above documents one such shift directly). Treat the shape of the finding, a large informational/transactional gap, as more durable than the exact 15% and 92% figures, and expect both to have moved by the time this guide is next reviewed.

AI recommendation sits closer to CHOSEN than to FOUND — and that changes what “optimising for AI search” should mean

This site’s framework for local business growth is FOUND → TRUSTED → CHOSEN: can a customer find the business, do they believe what they find, can they act on it right now.

Most AI-search advice implicitly treats AI recommendation as a new way to be found — an extra search engine to rank in. The evidence in this guide says that’s the wrong stage to file it under.

A 30× selectivity gap between what ChatGPT recommends and what the local pack shows means AI recommendation behaves like a harder, later filter, not an earlier, broader one.

Getting a Business Profile into local-pack shape earns a place among the field of businesses Google surfaces. It does not automatically earn a mention from an assistant choosing which one or two businesses to name out of everything it retrieved. That’s a CHOSEN-stage behaviour — a narrowing decision, not a discovery one — happening earlier in the funnel than it used to.

The consequence is that the tactics which actually move AI citation, per the evidence above, look more like CHOSEN-stage work than FOUND-stage work. Being cited favourably on an expert “best of” list, having a strong presence on industry-relevant domains, and carrying authoritative third-party reviews are all closer to reputation and trust signals than to the technical discoverability work — schema, crawlability, indexing — that dominates conventional SEO.

Optimising for AI search, done correctly on this evidence, looks a lot more like the reputation-and-proof work this site already argues most local businesses under-invest in relative to their spend on being found, and a lot less like a new technical checklist.

That’s the genuinely non-obvious implication. A business that has been pouring budget into FOUND-stage local SEO while treating reputation-building as an afterthought is, on this evidence, also under-prepared for AI search specifically — not because AI search is a separate discipline requiring separate technical work, but because it draws disproportionately on the CHOSEN-stage signals that discipline has been skipping.

Adoption is rising fast. Trust is a genuinely split verdict, not a foregone one

Self-reported adoption of a new behaviour tends to run high, because people round “tried it once” up to “use it.”

BrightLocal’s 2026 Local Consumer Review Survey found 45% of US consumers say they’ve used ChatGPT or another AI tool for local business recommendations, up from 6% the year before — a near-eightfold jump that deserves scepticism about the precise number even though the direction is credible. AI ranked third among sources consumers cited for local recommendations, behind Google (71%) and Facebook.

Trust tells a different, more contested story. The same survey found 40% of consumers trust AI platforms to provide business recommendations, against 32% who don’t — a genuine split, not the runaway confidence AI-adoption coverage often implies, with the remainder presumably undecided.

A further 42% say they trust AI “as much as traditional reviews.” And of the 82% who say they read AI-generated review summaries, only 23% would rely on the summary alone without checking star ratings or full reviews; 59% treat it as a starting point before cross-referencing, which sits consistently with what the wider review research says about how people verify.

The gap between adoption and trust is the more useful number here than either figure alone. Forty-five per cent having tried AI for a local search describes exposure. Forty per cent trusting the result, against thirty-two who don’t, describes something closer to actual influence on a decision — and it’s a materially smaller claim than “AI recommendations are replacing search.”

It gets more interesting once you separate general-population trust from the trust of people who already use these tools regularly. Yext’s 2026 Consumer Search Behaviors Report found 75% of existing AI users rate their trust in AI-generated local recommendations at 4 or 5 out of 5.

That is much higher than BrightLocal’s 40% general-population figure, because Yext asked only the self-selected subgroup who already use AI for search. It’s not a contradiction between the two studies; it’s a textbook selection-bias effect, and the two numbers should never be quoted against each other as though they measured the same thing.

More striking still: a Fractl and Search Engine Land study, repeating an identical question year over year, found the share of US consumers who consider AI search more helpful than traditional search fell from 82% to 54% between 2025 and 2026, a 28-point drop, while the share calling it actively less helpful rose from 3% to 17% — nearly a sixfold increase in outright scepticism in twelve months.

Adoption of AI for search rose through 2026 on other measures at the same time trust in it, on this measure, fell. Those aren’t contradictory findings about the same population. They describe different things — how many people have tried it, and how confident the ones who have are becoming — and both can be true simultaneously.

What this settles, practically: only 5% of AI users, per Yext, proceed straight from an AI answer to a purchase or booking without checking further. Fifty-three per cent verify through a separate Google or Bing search first; 49% visit the business’s own website directly before acting.

An AI assistant naming a business is closer to a FOUND-stage event than a TRUSTED or CHOSEN one — it puts the business in front of a customer, but the customer’s own trust check still happens afterward, on the business’s website and reviews, largely as it always did.

A business whose website can’t survive that follow-up check turns an AI mention into a wasted opportunity rather than a converted one. It is the AI-search version of a failure this site’s research keeps finding at the CHOSEN stage generally, regardless of channel.

A working framework: what to actually do, grounded in the findings above

This is a synthesis, not a generic checklist — every item below is tied to a specific finding earlier in this guide, not offered on its own authority.

Keep the Google Business Profile in the same disciplined shape it’s always needed, because it’s now feeding Gemini too. Gemini’s 100% accuracy figure and its comparative generosity (11% recommendation rate, the highest of the three assistants tested) both trace directly to its native grounding in Google Maps and Business Profile data — the exact input a business already controls for the local pack. This is the one AI platform where existing local-pack discipline transfers directly.

Treat the Yelp profile as a priority, not an afterthought, given the confirmed OpenAI data partnership. Complete business details, active review management, and accurate hours and contact information on Yelp specifically now feed ChatGPT’s answers directly, and plausibly influence Perplexity’s, per the sourcing section above. This is one of the few concrete, evidence-backed actions in this entire category.

Redirect schema-and-structured-data effort toward earning third-party mentions instead. The Ahrefs controlled study and Google’s own documentation both say schema is not the AI-citation lever it’s commonly sold as. Whitespark’s expert panel scored “best of” list presence, industry-domain prominence, and third-party review authority well above anything on-page. Pursue press coverage, industry directories, and genuine local publication mentions with the budget that might otherwise go to markup that a controlled test found does not move AI citation.

Build out informational content, not just transactional pages. The 15%-versus-92% AI Overview trigger gap means the content most likely to be intercepted by an AI answer before a searcher reaches the site is the how-to and explainer layer, not the “near me” and service pages most local SEO effort targets first.

Don’t block AI crawlers. This remains close to the only unambiguous technical action available: exclusion from a crawl makes inclusion in an AI-generated answer effectively impossible, regardless of anything else done well. Check what the site’s robots.txt actually disallows — Google’s crawler documentation sets out what a token such as Google-Extended does and does not control, and it is easy to have blocked more than was intended.

Check what major assistants currently say about the business, specifically, and correct what’s wrong where a correction path exists. ChatGPT and Perplexity’s 68% accuracy figure means roughly one in three details they surface about a given business may be stale or wrong. There’s no guarantee a correction propagates quickly, or at all, but leaving it unchecked guarantees nothing changes. The growth diagnostic is one way to get that read alongside the rest of the picture.

Don’t let an AI mention substitute for the trust-building work that follows it. Only 5% of AI users act on an AI answer without checking further; roughly half re-verify through a conventional search and roughly half go to the business’s own website. A site or review profile that can’t survive that follow-up turns a hard-won AI mention into a wasted one.

Don’t assume Google local-pack success transfers. Fewer than half of local-pack leaders also appear in AI recommendations for the same query. Treat AI visibility as a fourth surface to build deliberately, not a side-effect of ranking well elsewhere.

What’s still genuinely unknown, and what to avoid

Some gaps are worth naming rather than papering over. No substantial published study directly testing whether third-party mentions causally influence AI recommendation specifically for local businesses was identified in the sources reviewed on 8 September 2026. It’s a reasonable hypothesis, strongly consistent with Whitespark’s expert-panel scoring, but not yet a rigorously demonstrated finding in its own right.

Whether independent, single-location businesses face anything like the same 30× AI-versus-local-pack selectivity gap SOCi measured in multi-location brands is also unmeasured. And no official documentation from Perplexity about its own sourcing and weighting was identified, which means everything said about it here rests on observed output patterns rather than a disclosed policy.

Two things are worth actively avoiding regardless. The first is “GEO hacks” promising guaranteed AI citation. No credible, publicly verified method for reliably manipulating AI recommendation systems exists, and the schema example in this guide is a case study in exactly how a plausible-sounding tactic can turn out to do nothing when actually tested.

The second is assuming AI visibility follows automatically from good conventional local SEO. The evidence throughout this guide says that specifically isn’t reliable, and treating it as though it were is the single most common mistake this category currently produces.

This entire field is exactly where this site’s own research programme stays most active — see the AI local search statistics for the full, continually reviewed dataset behind this guide, and the local SEO statistics for how this fits inside the broader picture of what actually moves local visibility. Expect this guide to update as the evidence does, because on the numbers above, it will need to.

Questions

Is optimising for AI search different from regular local SEO?

It overlaps substantially but isn't the same discipline. AI assistants weight signals differently from Google's local pack: fewer than half of the businesses leading Google's local rankings also appear in AI-generated recommendations for the same query, and ChatGPT recommends roughly thirty times fewer businesses than the local pack shows, per SOCi's Local Visibility Index 2026. Ranking well on Google doesn't transfer automatically to AI visibility, so it needs treating as an additional surface rather than an assumed side-effect of conventional local SEO.

Does schema markup help a business get cited by AI assistants like ChatGPT?

The evidence says no, or at best not meaningfully. A controlled Ahrefs study added JSON-LD schema to 1,885 pages already receiving heavy AI citation and tracked the change against 4,000 control pages: the result was a small statistically significant decline for Google AI Overviews and changes indistinguishable from zero for Google AI Mode and ChatGPT. Google's own Search Central documentation states directly that there's no special schema.org markup required for AI features. Schema is still worth implementing for other reasons — see our local business schema guide — but it isn't the AI-citation lever it's commonly sold as.

Which AI platform should a local business prioritise: ChatGPT, Perplexity or Gemini?

On the current evidence, Gemini is the most reliable to be recommended by and the most accurate once it does — it's grounded directly in Google Maps and Business Profile data, the same data a business already manages for the local pack, and SOCi measured 100% business-data accuracy for it against 68% for both ChatGPT and Perplexity. That said, Gemini recommended only 11% of locations in the same study, so none of the three should be treated as solved; a business should maintain Google Business Profile accuracy as the common input across all of this rather than picking one platform to chase.

Should a local business invest specifically in its Yelp presence because of AI search?

There's a genuine, evidence-based case for it. OpenAI's confirmed data-licensing partnership with Yelp feeds over 330 million Yelp reviews across more than 8 million business listings directly into ChatGPT's local answers, and Perplexity's outputs show a consistent, independently observed lean toward Yelp and TripAdvisor content even without a confirmed partnership. That makes a complete, accurate, actively managed Yelp profile a plausible single point of leverage across two of the three major assistants — worth treating as a priority for any business it applies to, while remembering the terms are non-exclusive and could extend to other platforms or change again.

What actually predicts whether an AI assistant recommends a business?

Being mentioned favourably somewhere other than the business's own website, according to Whitespark's 2026 expert-panel scoring of AI search visibility factors. The highest-scoring factors were presence on expert-curated 'best of' lists, a dedicated page per service, prominence on key industry-relevant domains, the quality of unstructured mentions with no link, and the authority of third-party review sites. Three of the top five are about being cited elsewhere, not about on-page technical work — a genuinely different priority list from most 'AI SEO' guides, which tend to lead with schema and structured data.

Do Google's AI Overviews actually appear for 'near me' style local searches?

Less often than the headline statistics suggest. A Whitespark case study testing 540 queries found AI Overviews triggered for narrowly local-intent searches like 'plumber near me' only about 15% of the time, against 92% for informational queries in the same trades and cities, such as 'how do I know if I have a slab leak.' A single average figure across query types, commonly cited as around 68%, hides that gap entirely. The AI Overview risk for local businesses is currently concentrated in informational content, not in the transactional pages most local SEO effort targets first.

If AI recommends my business, does the customer actually trust it enough to act?

Not automatically, and the data suggests real caution. BrightLocal's 2026 survey found 40% of US consumers trust AI platforms for business recommendations against 32% who don't — a genuine split, not runaway confidence. Separately, Yext found only 5% of AI users proceed straight from an AI answer to a purchase or booking; roughly half re-run the check through a conventional search engine and roughly half visit the business's own website first. An AI mention gets a business in front of a customer; it doesn't replace the trust-building step that decides what happens next.

Should I block or allow AI crawlers on my website?

For a local business wanting AI visibility, allowing recognised AI crawlers rather than blocking them by default is the consistent choice. Being excluded from a crawl makes inclusion in an AI-generated answer effectively impossible regardless of anything else done well, and there's no credible evidence that blocking crawlers protects a local business from anything it should actually be worried about.

Is trust in AI search recommendations rising or falling?

The direction is genuinely unsettled and, on the clearest year-over-year data available, moving the opposite way to most adoption coverage. A Fractl and Search Engine Land study repeating an identical question found the share of US consumers rating AI search as more helpful than traditional search fell from 82% to 54% between 2025 and 2026, while the share calling it actively less helpful rose from 3% to 17%. That sits alongside rising self-reported adoption, which is the point: more people are trying AI search, and a meaningful share of them are coming away less convinced, not more.

What's the single most useful thing a local business can do given how unsettled this evidence still is?

Keep the Google Business Profile and Yelp profile complete and accurate, because those feed the most reliably grounded of the AI assistants and the one confirmed data partnership; get named on genuine third-party sources — review sites, local publications, industry-relevant domains — rather than over-investing in on-page schema; and treat any AI mention as a FOUND-stage event that still needs a website and review profile capable of surviving the trust check that follows, because the evidence shows most AI users verify before acting rather than trusting the answer outright.