The local business owner's guide to showing up when someone asks AI for a recommendation in 2026

Key takeaways

• BrightLocal found the share of consumers asking AI for local business recommendations jumped from 6% in 2025 to 45% in 2026, but SOCi's Local Visibility Index found only 1.2% of locations actually get recommended by ChatGPT, versus 35.9% visibility in Google's traditional local 3-pack.
• ChatGPT builds local answers largely from structured business data (think Yelp and Google Business Profile), not from crawling the open web the way Google does; ask for a suburb instead of a city and it often names nobody at all.
• Reviews and NAP consistency matter more for AI than for classic local SEO, and roughly 1 in 3 ChatGPT or Perplexity answers about a given business contain a factual error, so accuracy work pays off directly.
• Reddit's share of ChatGPT citations collapsed from around 4% to under 0.5% in August 2026, so stop chasing Reddit threads for ChatGPT visibility specifically; that effort is better spent on YouTube and Facebook if your customers lean on Google AI Overviews instead.
• A modest, well-structured local site can get cited. Domain authority above 90 is not required. Clean schema, dedicated service pages, and recent reviews beat a big budget.

Why this matters now, not next year

A year ago, "AI recommends local businesses" was a thought experiment for most shop owners. It isn't anymore. BrightLocal's Local Consumer Review Survey tracked the share of consumers using AI to find local businesses climbing from 6% in 2025 to 45% in 2026. That's not a gradual creep, that's a behavior shift that happened inside a single year.

And yet almost nobody is capturing it. SOCi analyzed roughly 350,000 locations across 2,751 brands and found only 1.2% get recommended by ChatGPT, 7.4% by Perplexity, and 11% by Gemini, compared to 35.9% visibility in Google's classic local 3-pack. SOCi calls AI visibility "3 to 30x harder" than ranking in traditional local search. I'd put it differently: it's not harder, it's just a different game that almost nobody local has started playing yet. Which, if you're reading this, is actually good news for you.

How AI actually decides who to recommend

This is the part most advice skips, and it's the part that determines whether your effort goes anywhere.

ChatGPT does not crawl the web the way Google does for local queries. It leans on a stack of structured data, Bing's web index for supplementary context, and a credibility filter that favors sources like Yelp, BBB.org, and local media while skipping paywalled or unattributed pages. When ChatGPT names businesses with star ratings, those ratings usually match the business's Google Business Profile. Ask the same city-level question twice and you'll get mostly the same list with different sentences around it, which tells you the list is grounded in a directory, not improvised.

There's also a popular claim floating around that Foursquare powers 60-70% of ChatGPT's local results. It's mostly wrong. That number traces back to a single May 2025 study of just 50 prompts in 5 Spanish cities, and it got broadened with every re-report until the caveats disappeared entirely. A rigorous 2026 re-test across 2,880 prompts and 4,607 runs in 12 US metros found Foursquare's actual citation share on ChatGPT's primary surface was 0.00%. Yelp occupied the structured business-card slot in 95.83% of runs instead, which lines up with a non-exclusive Yelp-OpenAI data deal reported in July 2026. TripAdvisor showed up 28.9% of the time, OpenTable 16.88%. So if you're deciding where to spend your limited hours this month, Yelp beats Foursquare by a wide margin, at least for ChatGPT.

One more honest finding worth internalizing: coverage thins out fast below the city level. Ask ChatGPT for "best electrician in Portland" and you get a list. Ask for your specific suburb and it often goes quiet. That's a gap in the assistants' current data coverage, not a verdict on your business, so don't panic if a hyper-local query returns nothing.

The six signals that actually move the needle

Pulled together from multiple independent analyses of how ChatGPT surfaces local businesses, the signals that consistently predict inclusion are:

1. Entity recognition. Your name, location, category, and contact info need to read identically everywhere AI might encounter them.
2. Review volume and sentiment, spread across platforms, not stacked on one.
3. NAP consistency. Name, address, phone number matching exactly across every directory listing.
4. LocalBusiness schema markup (JSON-LD) on your site.
5. Topical depth, meaning dedicated service pages and FAQ content, not one thin homepage trying to cover everything.
6. Third-party editorial mentions, like a local "best of" roundup, a chamber of commerce page, or a niche blog.

Notice how little of this is about your website's domain authority. Promptwatch's data on citation share by domain rank found mid-authority domains (DR 46-75) earned 46% of ChatGPT's citations in August 2026, while the top tier (DR 91-100) fell to just 3-4%. Sites with DR 0-30 still captured about 14%. Promptwatch's own analysis backs this up directly, and the practical read for a plumber or a dentist is simple: you do not need to outrank Forbes. You need to be structured well enough that AI can trust what it finds.

What AI actually cites, by content type

Promptwatch's content-type data from late September/early October 2026 found product pages are the single most-cited format on ChatGPT overall (22.4%), with landing pages close behind (16.8%). For commercial, "best X near me"-style prompts specifically, product pages, landing pages, and listicles together make up roughly 70% of citations. Reviews only climb in importance once your business is already named in a branded prompt, where review citations jump to 5.6% of the mix, about three times the baseline. The lesson: get the service page right first, because that's what earns you the initial mention. Reviews then do the work of reinforcing it.

Fix your directory and listing foundation first

Before touching your website, audit the basics, because inconsistency here actively reduces AI's confidence in recommending you, more than it hurts traditional Google ranking:

• Google Business Profile, fully filled out, categories correct, hours current.
• Bing Places. ChatGPT draws on Bing's index, so this one is not optional the way some owners assume.
• Yelp, given the data above showing it occupies the structured business-card slot in over 95% of ChatGPT's local answers.
• Apple Business Connect and any industry-specific directory relevant to your trade.

BrightLocal's analysis of 1.9 million local AI citations found Google Business Profile accounts for 28.5% of all local AI citation sources, and Yelp appears as a source in 80% of all ChatGPT local answers. That's about as close to a must-do list as this space gets.

Build pages AI can actually extract facts from

AI models process natural language, not keyword density. A page that clearly answers who you are, what you do, where you're located, and why someone should pick you, in plain conversational English, is easier for an AI to lift facts from than a page stuffed with search terms. A well-structured FAQ section on each service page helps enormously here, because it lets the model pull a direct answer instead of having to infer one.

Add LocalBusiness schema (JSON-LD) to every location and service page. Schema.org's LocalBusiness type supports fields like aggregateRating, review, priceRange, openingHoursSpecification, and sameAs, which lets you link out to your Yelp and Facebook profiles for entity verification. Google's own documentation recommends including geo coordinates at 5+ decimal places and a properly formatted phone number with country and area code. This is unglamorous work, but it's exactly the kind of structured signal that AI systems lean on when deciding whether to trust a listing.

Don't accidentally block the AI crawlers you're trying to impress

This one catches a surprising number of business owners off guard. On July 1, 2025, Cloudflare quietly flipped its default policy for new domains from "AI can access your content" to "AI is blocked unless you opt in." Cloudflare powers roughly one in five websites. If your site sits on Cloudflare and nobody has touched the AI Crawl Control settings since then, there's a real chance you're invisible to the very crawlers you're trying to court.

It got more complicated in September 2026, when Cloudflare changed what "Block" means. A blanket Block setting now also catches mixed-use crawlers including Googlebot, Bingbot, and Applebot, so a well-intentioned attempt to keep AI training bots out can accidentally cut off your normal search indexing too. The fix is to use the "Disallow AI Training" setting specifically rather than a blanket Block, and to explicitly allow PerplexityBot, OAI-SearchBot (ChatGPT's retrieval crawler), and ClaudeBot if AI visibility matters to your business. Check this today. It takes five minutes and it's the kind of thing that silently costs businesses visibility for over a year without anyone noticing.

Where social content actually helps (and where it's a waste of time)

Here's a genuinely useful piece of data that contradicts a lot of generic advice floating around: Promptwatch's October 2026 analysis of social media citations by AI model found ChatGPT has nearly stopped citing social platforms altogether, just 0.46% of its citations combined across all seven tracked platforms, down from 5.8% previously. Reddit specifically collapsed from about 5.19% to 0.09% after an August 14, 2026 drop. If your plan was "get mentioned in local subreddit threads to show up in ChatGPT," that plan stopped working months ago, documented in Promptwatch's data on Reddit citations dropping in ChatGPT.

Google AI Overviews is the opposite story. Social platforms make up 12.2% of its citations, about 26 times ChatGPT's share, with YouTube leading at 4.28% and Facebook at 2.38%. So the right move depends on which AI engine your customers actually use. If you serve a market where Google AI Overviews shows up constantly in local search, a YouTube walkthrough of your shop or a Facebook presence with recent, active posts is worth real time. If ChatGPT is the dominant source, that same effort is close to wasted.

Review strategy: recency beats raw volume

Whitespark's 2026 local search ranking factors survey, which polls local SEO experts rather than measuring AI directly, found review recency now outweighs raw review count. A business with 10,000 historical reviews and total silence for the past year can actually rank worse than a business with fewer reviews arriving steadily. This lines up with how AI reads reviews too: it's extracting sentiment and specific mentions (wait times, pricing, cleanliness, expertise), not just averaging stars. A business with 200 recent reviews spread across Yelp, Google, and Facebook sends a stronger signal than 50 reviews sitting only on Google.

Practically, that means building a habit of asking satisfied customers for a review right after service, on more than one platform, rather than running one big review push and letting it go quiet.

Tracking whether any of this is working

You can't improve what you can't see, and local AI answers change week to week, so a one-time check tells you almost nothing. A few options at different budget levels:

Free, manual: literally type your customer's question into ChatGPT and Google's AI yourself every week or two and note who gets named. Slow, but free and honest.

Promptwatch tracks how your business is cited across ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews, and AI Mode, with crawler logs that show exactly which pages AI systems are actually reading on your site and whether they hit errors. For a local business juggling multiple locations or service areas, that visibility into what's working, rather than just whether you got mentioned, is the difference between guessing and knowing.

Promptwatch

Track and optimize your brand's visibility in AI search engines

View more

For smaller budgets, Otterly.AI starts at $29/month for 15 prompts across four engines, and BrightLocal bundles a Local AI Visibility Tracker starting at $31/month as an add-on to their existing local SEO suite, which is a reasonable choice if you already use BrightLocal for citations and rankings.

If you want to browse more options before committing, the GEO software directory at bestgeosoftware.com lists a broader range of monitoring tools by feature and price point.

A realistic 30-day plan

Week one: claim and fully complete Google Business Profile, Bing Places, Yelp, and Apple Business Connect. Fix any NAP mismatches you find.

Week two: check your Cloudflare AI Crawl Control settings (or ask whoever manages your hosting to check). Add LocalBusiness schema to your homepage and top service pages.

Week three: write or rewrite one dedicated page per service you offer, in plain language, with a genuine FAQ section addressing the questions customers actually ask you on the phone.

Week four: start a lightweight, ongoing review request habit across at least two platforms, and run your first set of "what does AI say about me" test queries, either manually or with a tracking tool, so you have a baseline to measure against next month.

None of this requires a marketing department. It requires about an afternoon a week for a month, and then an ongoing habit of checking in. Given that 45% of consumers are already asking AI where to go, and only 1.2% of local businesses are showing up when they do, the businesses that bother to do this basic work in 2026 have an unusually wide opening in front of them.