Straight answer: ChatGPT and Perplexity mention small businesses that have specific, quotable facts on their pages: real numbers, named examples, clear answers to exact questions. They skip pages full of soft opinions and marketing fluff, no matter how well those pages rank on Google. I ran a fifty-prompt test on my own site to check, and the result changed what I write.
I Ran The Test On My Own Site First
In January 2026 I sat down with a spreadsheet and forty three prompts, the kind of things a real person types when they're stuck: "how much does an AI consultant cost for a small business", "best way to use AI for email marketing UK", "how do I get my small business found by AI search". I ran each one through ChatGPT and Perplexity and wrote down whether my site got mentioned.
Out of forty three prompts, I was mentioned or cited four times. Four. Out of forty three. I've been writing marketing content for over a decade, I have a Forbes listing, I have hundreds of blog posts. Four mentions.
That stung more than I expected, so I dug into which four posts made the cut and which forty odd didn't. The pattern was obvious once I saw it laid out: the posts that got cited had a number, a named company, or a step-by-step in the first two hundred words. The posts that got skipped opened with a nice friendly paragraph about "the changing landscape of marketing" and took four hundred words to say anything specific.
What Gets Skipped Every Single Time
The AI models aren't being unkind. They're pattern matching for a quotable sentence they can lift and attribute. If your opening paragraph is warm and vague, there's nothing to lift. Here's what I found consistently ignored, across my own site and about a dozen client sites I checked afterwards:
- Posts that open with a question and then spend three paragraphs building up to the answer
- "Ultimate guide" style pages with no real numbers, only phrases like "many businesses" or "a growing number of companies"
- Pages where the useful fact is buried under an SEO-friendly 150-word introduction written for Google, not for a person or a machine reading it
- Anything that reads like it was written to hit a word count rather than answer a question
Compare that to how Ahrefs built its content authority. Their blog posts open with the number, the study, the specific finding, then explain it. That's exactly the shape AI models want to pull from. It's not an accident that a data-led SEO company built content that plays well with AI search too, because good clear specific writing was always the right call, it's just that now there's a second reader checking your work.
Five Things I Changed On My Own Pages
After the spreadsheet exercise I went back through my ten best performing posts and rewrote the openings. Not the whole post, just the first two hundred words. Here's exactly what I did, in order:
- Moved the number to sentence one. Not "AI consultants can be expensive" but "AI consultants in the UK typically charge £500 to £2,000 a day, or £3,000 to £15,000 for a short project." A machine can quote that. It cannot quote "can be expensive."
- Named the source of every claim. If I say something costs a certain amount, I say who charges that and why, or where the figure came from.
- Added a bolded one-paragraph answer at the very top of the post, before any scene-setting, so there's a self-contained quotable chunk sitting right at the start (you're reading one now).
- Wrote a short FAQ at the bottom with real questions, phrased the way people type them into a search bar or a chatbot, not the way a marketer would phrase a headline.
- Cut every sentence that started with "In today's fast-paced world" or its cousins. Those sentences carry zero information and machines skip straight past them, same as a busy human reader would.
I reran the same forty three prompts six weeks later. Mentions went from four to eleven. Not because I wrote more, I wrote less overall, I just moved the useful part to the front and stripped the padding.
The Kind Of Page That Gets Quoted On Purpose
Interactive tools and calculators get cited more than almost anything else, because they produce a specific number for a specific input, which is exactly what these models are hunting for. I wrote about this shift a while back when I looked at how interactive calculators pull in and convert visitors, and the same logic applies here even harder now. A calculator that tells someone "based on your inputs, you'd pay roughly £1,200" is far easier for a chatbot to reference than a paragraph explaining pricing philosophy.
Brand storytelling still matters too, just differently. Look at how Disney's marketing has been built around specific, repeatable facts people can quote back, or how GoDaddy anchored its brand around one clear message for years rather than trying to say everything at once. Same goes for Headspace's simple, specific positioning. None of these brands write vague. They pick one clear thing and say it plainly, over and over, with a number or a name attached. That's the same instinct small business owners need for AI visibility, just applied to a blog post instead of a whole brand.
Does Any Of This Make You Money?
Here's the bit most people writing about "AI search optimisation" won't tell you straight: being mentioned by ChatGPT does not send you traffic the way a Google ranking does. When ChatGPT quotes your page, most people never click through. They read the answer, they close the app, they move on. I've had client sites get quoted repeatedly in Perplexity results and see almost no measurable increase in visits.
So why bother? Two reasons, and only two. First, brand recall. If ChatGPT keeps naming your business when people ask about AI consultants in your city, or about email marketing costs, that name sticks even without a click, the same way hearing a brand mentioned on the radio sticks even though you can't click a radio ad. Second, and this is the bigger one for me, the discipline of writing this way makes every piece of content better for humans too. Clearer openings, real numbers, no padding. That's good writing whether a machine reads it first or not.
I'd be lying if I said chasing AI citations is a growth channel on its own right now, in 2026. It isn't, not yet, not in any way I can measure in invoices. Treat it as a free byproduct of writing clear content, not as a strategy you build a quarter around.
There's a behavioural reason we get so excited about this anyway, and it's worth naming. Richard Thaler's work on why we overweight visible, immediate signals explains a good chunk of why business owners obsess over an AI mention screenshot more than a steady 3% month on month traffic climb. The screenshot feels like proof. The traffic graph feels boring. The traffic graph is the one that pays your rent.
A Ten Minute Weekly Check You Can Do Yourself
You don't need a tool or a subscription for this. Every Monday, pick five questions your ideal client would type into a chatbot, the same ones you'd expect them to search on Google. Type each into ChatGPT and into Perplexity. Note whether your business, or a competitor's, gets named. Write down the exact phrasing that got cited if you spot one. Over a month you'll have twenty data points and a clear pattern of what kind of content earns a mention in your specific niche. It costs you forty minutes a month and nothing else.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
If you run a small team and don't have forty minutes a month spare because you're drowning in the actual work of the business, this is exactly the sort of gap I help clients close when I work as an AI consultant for small business, auditing what's already on a site and rebuilding the openings so they carry real information instead of padding.
Where This Sits Next To Normal SEO
None of this replaces proper SEO work. Keyword research, backlinks, technical site speed, all of that still decides whether Google sends you traffic, and Google traffic still converts far better than an AI mention does. Think of AI visibility as a layer on top, not a replacement. Write for Google's ranking factors first, then tighten the opening two hundred words of your best pages so a machine skimming for a quotable fact finds one immediately. Both jobs get done by the same edit.
Frequently asked questions
Does getting mentioned by ChatGPT bring customers?
Rarely in a way you can trace directly. Most people who see your business named in a ChatGPT answer never click through to your site. The value is brand recall and the discipline of clearer writing, not a measurable traffic spike, so don't build a content plan around it alone.
How do I check if my business is already mentioned by AI chatbots?
Type the exact questions your customers would ask into ChatGPT and Perplexity, the same phrasing they'd use, and read the answers carefully. Do this with five to ten real questions rather than one, since results vary a lot by phrasing.
What's the fastest change to make to an existing blog post?
Rewrite the first two hundred words so the specific answer, number, or fact appears in sentence one, before any scene-setting. Add a short FAQ at the bottom using real question phrasing. That single edit moved my own citation rate from four out of forty three prompts to eleven.
Should small businesses hire someone specifically for AI search visibility?
Not as a standalone service, no, it's too early and too unmeasurable for that spend to make sense on its own. It's worth folding into a wider content or AI marketing review, where clearer, more specific writing pays off across Google, social, and AI answers at the same time.
Related reading: I Asked AI to Audit My Own Blog. It Told Me to Delete Half of It. and I Write Client Proposals With AI in 20 Minutes. Here's the Bit Nobody Tells You.
This builds on my main AI marketing guide, my main guide on the topic.
Want this done for you? See write for us about small business.
