How AI Decides Which Businesses to Recommend: We Tested It
I’ll be honest. When people ask me how AI decides which businesses to recommend, my answer used to be “nobody knows for sure”. It’s true, but it’s a bit of a cop-out, and I don’t love cop-outs.
So we ran our own experiment. We drew a service out of a hat (it came out business advisers), asked ChatGPT and Google AI Mode who to call in Sydney, Newcastle and Melbourne, and then went through the websites of the firms that kept getting named and the ones that never did.
In short: the businesses AI recommended tended to have three things in common: named case studies, a clear audience and real team profiles on their own websites. The ones it ignored were much less likely to have all three.
If you’ve already wondered why your business isn’t showing up in AI search, this is the evidence behind it. The service doesn’t matter. We’re running the same test on other services and other types of business, and I’ll share those as we go. What I wanted was the pattern underneath. This is one industry and a handful of questions, so treat it as a guide, not a guarantee. Different industries and different search terms will behave differently.
Which businesses does AI recommend: big firms or small ones?
Both. The big national names showed up, but they didn’t take over. Small and mid-sized firms made up around six in ten of the mentions, particularly when the question was specific. Ask about selling a business in three years and the specialists came through.
So size wasn’t the deciding factor. Something else was.
What do the businesses AI recommends have in common?
Their websites were clearer and better backed up. We scored ten websites out of 10. Five were firms AI named again and again. Five were similar firms, matched on city, size and services, that never appeared once. The first group averaged 7.8. The second averaged 5.0.
| Website signal | Named by AI | Never named |
|---|---|---|
| Overall score (out of 10) | 7.8 | 5.0 |
| Case studies naming the client and result | 4 of 5 firms | 1 of 5 firms |
| Clear about who they serve (out of 2) | 1.6 | 1.0 |
| Real team profiles (out of 2) | 1.6 | 1.0 |
Most of that gap came down to three things.
Case studies: proof AI can repeat
Four of the five firms AI named had case studies with a real client, a real problem and a real result. When we read the reasons AI gave for its picks, it pointed to those case studies. Think about it from AI’s side. A list of services gives it nothing to say about you that it couldn’t say about anyone.
A clear audience: who your business is for
The named firms said it plainly, often on the homepage. Several of the ones AI ignored talked about serving “diverse industries” with “tailored solutions”. That might be true, but when a buyer asks for help with one particular kind of business, AI has nothing to match it to.
Team profiles: the people behind the work
Bios with real experience and specialities beat a grid of names and job titles. It’s the same thing a person looks for before they pick up the phone.
Do Google reviews affect whether AI recommends your business?
Possibly less than you’d expect, but one small test can’t tell us how much weight either tool gives them. One business with a perfect rating never got named, and one of the most-named had 3.7. Reviews still build trust with people, and in our test a good rating didn’t make up for a thin website.
Do ChatGPT and Google AI Mode recommend the same businesses?
Often not. Of the 75 or so firms we logged, only 21 turned up in both tools. If you check one and you’re there, don’t assume you’re in the other.
Both tools often linked straight to the firm’s own website, which tells you where to look first.
How to make your website easier for AI to recommend
Most business websites were built before anyone thought to ask AI for a recommendation. They describe what the business does, and a lot of them stop there. Who it’s for, who did the work and what happened for the last client often aren’t on the page.
AI can only repeat what you’ve published. If the proof, the people and the “who we help” aren’t there, it has little to work with, however good you are. The encouraging part is that these are specific gaps, and they can be fixed. (For the longer story on why AI skips some websites, read our guide to getting your business found in AI search.) Here’s what to look for on your own site:
Show who you help. Put one plain sentence on your homepage that says who your business is for.
Publish proof of your work. Add a case study that names the client and the result, on its own page.
Introduce the people behind the business. Give each person a short bio with their experience and what they’re good at.
Want to see what AI says about your business? Book a free AI website audit
This is the work we do every day, and it’s why we ran the test. Book a free AI website audit and we’ll go through your website the way AI would. You’ll see how it’s likely to describe you, what’s missing and what to fix first.
We can’t promise AI will name you, and nobody honestly can. We can make sure it has what it needs. If the audit shows your website is what’s holding you back, we can talk about building one around your industry, your buyers and what the audit found.
Book your free audit →
How we tested AI business recommendations, and what the test can’t tell you
We asked 12 questions in each tool on 4 November 2026, in a fresh chat every time, which gave us 24 answers and around 135 business mentions. We haven’t named any of the businesses, because the point isn’t who won.
To compare websites, we picked five firms AI named repeatedly. Then we picked five similar firms that never appeared, matched as closely as we could on city, size and services. We scored each website from 0 to 2 on five signals:
- Who it’s for: does the site say plainly who the business helps?
- Named proof: are there case studies or results that name the client?
- Team profiles: do the bios show real experience, or only names and job titles?
- Consistent story: does the business describe itself the same way across its website and profiles?
- Reviews and listings: is there a Google Business Profile with reviews, and other listings?
We’re re-running the test and trying it on other services too, and we’ll share what we find here. We’re re-running experiments regularly and will continue to let our clients know the findings as they come.
About Deb
Meet the author
Deb Croucher is the founder of Brilliant Digital. She works directly with SME owners to create strategy-led, full-service marketing that delivers, not just in Google, but in the new world of AI-powered discovery.
You Might Also Like…
How AI Decides Which Businesses to Recommend: We Tested It
We drew a service out of a hat and asked AI who to hire. Then we read the websites of the winners and the also-rans. The difference was easy to spot.
Why isn’t my business showing up in AI search?
You rank on Google, but AI barely knows you exist. Here’s what your website isn’t saying, and what AI looks for before it recommends a business.
Outsource digital marketing or hire in-house?
One marketing salary rarely buys a complete marketing team. Discover what in-house capability really costs, and where outsourcing can cut waste without cutting growth.
Websites have a used by date SMEs must evolve
Your business doesn't stand still. Your customers don't stand still. Search doesn't stand still. So why would your website?
The curse of the quick click
What happens when the promise of quick clicks and a high Google ranking erodes long-term marketing strategies, and with little result.
Analytics: The foundation of B2B marketing growth
Often overlooked, analytics and reporting are essential to B2B marketing success, helping businesses optimise campaigns and drive growth.
It’s not a marketing problem; It’s a strategy problem
For many B2B and SME businesses, the issue is not the marketing; it’s the strategy behind it that's why the marketing isn't working as it should.
Why does an agency matter for your business marketing?
When your business relies on digital marketing, engaging a digital marketing agency is a wise commercial decision. Less overhead, maximum return.
Search trust is the key to success
The jury is no longer out, and the stats cannot be ignored. You must be a trusted source of information, or you’re invisible in online search
AI vs custom websites: Trust always wins
Why built-for-scale AI websites can’t replace custom WordPress builds with human content that drives SEO, AEO visibility and real customer trust.
Let’s Grow Your Business Together
Key Takeaways
FAQs
How does AI decide which businesses to recommend?
In our test, the businesses AI named tended to publish named case studies, say clearly who they serve and show their team’s experience on their own website. It’s one industry and a handful of prompts, so it’s a pattern, not a rule.
Do I need case studies to be recommended by AI?
There’s no guarantee, but four of the five businesses AI named repeatedly had case studies naming the client and the result. Only one of the five it ignored did.
What makes a good case study for AI search?
Name the client (with their permission), describe the problem, say what you did and give a result a stranger could understand. Put it on its own web page, not in a PDF, so AI can read it and link to it.
Can a small business get recommended by AI over bigger competitors?
Yes. Small and mid-sized firms took around six in ten of the mentions in our test, and did best when the question was specific and their website clearly matched it.
Why do ChatGPT and Google AI Mode recommend different businesses?
They draw on different sources and don’t agree often. In our test, only 21 of about 75 firms appeared in both, so it’s worth checking each tool separately.
Do I need a new website to be recommended by AI?
Sometimes. Some businesses need clearer pages and stronger proof, and others have outgrown their site. A free website audit tells you which before anyone suggests a rebuild.