I Tested How AI Search Sees 80+ Small Businesses — Here Are the Patterns
Over the past three weeks, we asked ChatGPT, Perplexity, and Google AI Overviews to recommend businesses across 21 industries in 13 countries. We tracked which businesses appeared, which didn't, and what the recommended ones had in common. Six patterns kept showing up.
In this article
How we tested: methodology
Between late February and mid-March 2026, we ran a structured test. For each of the 80+ businesses in our sample, we asked three AI search engines the same question: "Recommend a good [industry] in [city]."
We tested across ChatGPT (GPT-4), Perplexity, and Google AI Overviews. We covered 21 industries — from restaurants and hotels to dental practices, law firms, wedding venues, and auto repair shops. The businesses spanned 13 countries including the UK, US, Canada, Australia, France, Germany, Ireland, Sweden, South Africa, and more.
Every business in our sample was a real, operating SME with an active online presence. None were digital-first companies, SaaS startups, or e-commerce pure-plays. These are the kind of businesses that depend on local or regional customers finding them — exactly the businesses most affected by the shift to AI search.
Test scope at a glance
80+
Businesses tested
21
Industries covered
13
Countries
3
AI platforms
The headline numbers
The results were stark. When we asked AI to recommend a business in a specific industry and city, the vast majority of the businesses we tracked did not appear in the results at all.
of businesses in our sample were completely invisible to at least one AI search engine — not mentioned at all, even in a list of alternatives
were invisible across all three platforms — ChatGPT, Perplexity, and Google AI Overviews had no idea they existed
were consistently recommended across all three AI platforms — named, described, and presented as a top choice
That 12% — the businesses AI consistently recommends — shared six clear patterns. These aren't opinions or guesses. They're what the data showed, again and again, across industries and countries.
Third-party mentions are the #1 signal
This was the single strongest predictor. Businesses that appeared in AI recommendations were mentioned on websites other than their own: local guides, industry directories, news articles, blog features, and “best of” lists.
A bakery in Melbourne that was featured in a local food blog and two city guides? ChatGPT named it. A bakery down the street with a better website but zero external mentions? Invisible. Your own website is necessary but not sufficient. AI needs to hear about you from other sources to trust that you're worth recommending.
We saw this hold true across every industry and geography we tested. A veterinary practice in Edinburgh, a wedding photographer in Berlin, a chiropractic clinic in Calgary — the ones AI recommended had one thing in common: other websites talked about them.
What to do about it:
- •Get listed in every relevant industry directory for your area — not just Google, but niche directories your industry uses
- •Pitch a story, quote, or feature to a local publication or trade blog. Even a single mention can shift AI visibility
- •Partner with complementary businesses for cross-mentions (a florist linking to a wedding venue and vice versa)
- •Target: Aim for at least 5 third-party sources that mention your business by name
Reviews need volume AND platform diversity
Having reviews helps. But the businesses AI consistently recommended didn't just have Google reviews — they had reviews across multiple platforms. Google, TripAdvisor, Yelp, Trustpilot, industry-specific platforms (Avvo for lawyers, Zocdoc for doctors, Booking.com for hotels).
A hotel in Reykjavik with 200 Google reviews but nothing on Booking.com or TripAdvisor? AI was less confident. A smaller hotel with 80 reviews spread across three platforms? Named by ChatGPT every time.
This makes sense when you consider how AI models are trained. They synthesise information from many sources. If all your social proof lives on one platform, the AI has fewer data points to work with. Diversity of review sources signals genuine, broad-based reputation.
What to do about it:
- •Identify the top 3 review platforms for your industry and make sure you have an active, claimed profile on each
- •When asking customers for reviews, rotate which platform you point them to — don't always default to Google
- •Respond to reviews on every platform — AI reads responses too, and they add context about your business
- •Quick win: Detailed reviews that mention specific services matter more than generic “great service!” ratings
Niche specialists beat generalists every time
When we asked ChatGPT for “a good family lawyer in Auckland,” it didn't recommend the biggest law firm in the city. It recommended a boutique firm that specifically describes itself as a family law specialist. When we asked for “boutique yoga studios in Singapore,” it picked the studio with classes capped at 10 people and 15+ years of instructor experience — not the large chain.
AI search engines match queries to answers, not keywords. When someone asks for something specific, the AI looks for businesses that clearly demonstrate that specific expertise. A website that says “We do everything” gives AI nothing to work with. A website that says “We specialise in X for Y in Z” gives AI exactly what it needs.
This was especially pronounced in our cross-industry analysis. In every single industry we tested, the specialist outperformed the generalist in AI recommendations.
What to do about it:
- •Create dedicated pages for each service you offer, rather than one catch-all “Services” page
- •Include your location and speciality in page titles and headings: “Wedding Photography in Bristol” not “Our Work”
- •Write content that answers specific questions your ideal customers ask — the more specific, the better AI can match you to queries
- •Key insight: If you serve multiple niches, each one deserves its own page with its own depth — topical depth beats generic breadth
Structured data gives AI confidence
This one surprised us. We expected content and reviews to dominate. But businesses with proper structured data (schema markup) on their websites appeared in AI results at a noticeably higher rate — especially in Google AI Overviews.
Structured data is invisible code on your website that tells search engines exactly what your business is: name, type, location, opening hours, services, price range. It's like handing AI a neatly formatted fact sheet instead of making it read through paragraphs to figure out the basics.
A dental practice in Manchester with LocalBusiness and Dentist schema markup? AI knew its address, specialties, and opening hours — and recommended it confidently. A competing practice with a beautiful website but no structured data? AI wasn't sure what it was looking at.
What to do about it:
- •Add LocalBusiness schema markup to your website — most website builders (WordPress, Squarespace, Wix) have plugins or built-in tools for this
- •Include your industry-specific schema type (e.g., Restaurant, LegalService, DentalClinic, LodgingBusiness) for more precise AI matching
- •Add FAQ schema to your most-asked-questions page — this feeds directly into AI answer generation
- •Free test: Google's Rich Results Test can check if your website has structured data already — most SME sites don't
Freshness matters more than you think
We noticed a pattern with businesses that had “stale” online presences. If the last blog post was from 2023, the last Google Business update from 2022, and the most recent review from six months ago, AI was less likely to recommend them.
This makes intuitive sense. AI search engines want to recommend businesses that are actively operating and relevant right now. A website that hasn't been updated in two years could be a business that's closed. Recent reviews, recent content, and recent social activity all signal that you're alive, active, and worth recommending.
A landscape designer in Austin who posts project photos monthly and has a review from last week? ChatGPT named them. A competitor with a bigger portfolio but no activity since 2024? Skipped entirely.
What to do about it:
- •Post at least one piece of new content per month — a blog post, a case study, a project highlight, or even a seasonal update
- •Update your Google Business Profile regularly with photos, posts, and responses to reviews
- •Create a steady drip of reviews rather than a burst — 2 reviews per month beats 20 reviews once a year
- •Low effort: Even updating the copyright year in your website footer and refreshing your “About” page sends a freshness signal
The “story factor” — AI loves a narrative
This was the most unexpected finding. When AI recommended a business, it didn't just list the name. It told a story: “Founded in 2001, this family-run bakery in Melbourne's South Yarra is known for its sourdough and has become a neighbourhood institution.”
AI could tell this story because the information existed online — on the business's website, in reviews, in press coverage. Businesses with a clear narrative (founding story, mission, unique angle, history) gave AI the raw material to craft a compelling recommendation.
Businesses without a story? If AI mentioned them at all, the recommendation was flat: “[Business Name] is another option in the area.” That's not the kind of recommendation that wins a customer. As we noted in our analysis of why AI recommends your competitor, the businesses that win are the ones AI can speak about confidently and specifically.
What to do about it:
- •Write a compelling “About” page that tells your founding story, your mission, and what makes you different — in concrete terms
- •Include specific numbers and facts: years in business, number of clients served, team size, notable projects
- •Make sure your story appears in more than one place — your website, your Google Business description, your LinkedIn, and any directory profiles
- •Think of it this way: If a journalist had to write one sentence about your business, could they? That sentence is what AI will use
The AI visibility scorecard
Based on these six patterns, here's a quick self-assessment. Rate yourself honestly on each factor:
| Factor | Strong (AI likely recommends you) | Weak (AI likely ignores you) |
|---|---|---|
| Third-party mentions | 5+ independent websites mention you by name | Only your own website and social media |
| Review diversity | Reviews on 3+ platforms, detailed and specific | Reviews on 1 platform only, mostly generic |
| Niche clarity | Clear speciality, location, and audience on every page | Generic “we do everything” messaging |
| Structured data | Schema markup on key pages (LocalBusiness, FAQ, etc.) | No structured data on website |
| Freshness | New content or reviews within the last month | Last update was 6+ months ago |
| Story / narrative | Clear founding story, mission, and unique angle online | No story — just a service list and contact form |
If you're in the “weak” column on 3 or more of these, you're almost certainly invisible to AI search. The good news? Each factor is fixable. And you don't need to fix all six at once — even improving two or three can shift your visibility significantly.
Want a quick benchmark? Try our free AI visibility score tool — it takes 30 seconds and gives you an instant snapshot of where you stand.
Frequently asked questions
How representative is a sample of 80+ businesses?
It's not a peer-reviewed study — it's a practitioner's analysis. But the consistency of the patterns across 21 industries and 13 countries gives us high confidence. The same six factors kept emerging whether we were testing a florist in Stockholm or a law firm in Auckland. That cross-category consistency is hard to dismiss.
Do these patterns apply equally to ChatGPT, Perplexity, and Google AI Overviews?
Mostly, yes. The core factors (third-party mentions, reviews, niche clarity) mattered across all three. Structured data had the strongest effect on Google AI Overviews, which makes sense given Google's existing reliance on schema markup. Perplexity seemed to weight freshness more heavily. But no single factor was irrelevant on any platform. Our guide to spotting AI invisibility covers the platform-specific differences in more detail.
Can a small business with no budget improve AI visibility?
Absolutely. Most of the actions in this article are free. Claiming directory listings, asking customers for reviews on different platforms, updating your website's About page, adding structured data — none of these cost money. They cost time and consistency. Our beginner's guide to appearing in ChatGPT walks through the free steps in detail.
How quickly can I expect to see changes?
Quick fixes (unified business information, structured data, updating stale content) can show effects within weeks. Building third-party mentions and review diversity takes 1–3 months of consistent effort. The businesses that broke into the “consistently recommended” category in our test didn't get there overnight — but they all started with the basics and built from there.
Find out exactly where your business stands
Our AI PR + SEO Growth Guide analyses your visibility across ChatGPT, Perplexity, and Google AI Overviews, scores you on all six patterns, and gives you a prioritised action plan.
- Full AI visibility audit across all major platforms
- Scored against all 6 visibility patterns
- Competitor benchmarking and gap analysis
- Prioritised 30-day action plan + PDF report
One-time payment · Delivered in 48 hours · No subscription