The short version: Most small business owners have never typed their own company name into ChatGPT or Perplexity to see what comes back, and when they finally do, the answer is usually nothing. Getting cited by AI tools has very little to do with how well written your website is and a lot to do with whether other people are already talking about you somewhere else. This post shows you exactly how to run the audit yourself, what the results mean, and what to do when they’re bad.
Worth reading next: How to Use ChatGPT to Write a Resume That Gets You Interviews.
What an AI citation audit is
An AI citation audit is simply this: you ask ChatGPT, Perplexity, Claude, and Google’s AI Overviews a set of questions your ideal customer might ask, then you check whether your business, your name, or your content shows up in the answer. Not your Google ranking. Not your website traffic. Whether an AI tool, when asked a real question, chooses to mention you at all.
I started doing this for my own business in late 2025 because clients kept asking me the same thing: “will AI kill my SEO?” I didn’t have a confident answer, so I built a spreadsheet, wrote out 12 questions a potential client of mine might ask, and ran each one through four tools. That’s 48 checks. I did the same thing again three months later to see what had shifted.
The exact steps to run your own audit
- Write 10 to 15 questions your customer would type, not keywords. “Best AI consultant for a small business in Manchester” not “AI consultant Manchester.”
- Run each one in ChatGPT (with browsing/search on), Perplexity, Claude, and check the Google AI Overview if one appears.
- Record whether your business is named, whether a competitor is named instead, and whether any of your actual content is quoted or linked.
- Note the source the AI pulled from. This is the bit people skip and it’s the most useful column in the whole spreadsheet.
- Repeat the same set of questions every 8 to 12 weeks. One-off audits tell you almost nothing because these answers shift week to week.
That “source” column is where the real learning happens. When I was cited, it was almost never because of my blog. It was because of a Forbes Councils piece I’d contributed to, a podcast transcript, or a roundup article someone else had written that happened to mention me. Out of the times I appeared across those 48 checks, roughly two thirds traced back to something published on someone else’s site, not mine.
The uncomfortable part nobody wants to say out loud
Here’s the bit that most “how to win at AI search” content skips over because it’s not a fun thing to sell: if your business has no press, no podcast appearances, no guest posts, no directory listings, and no one else on the internet mentioning your name, you can rewrite your entire website with perfect structure and you still probably won’t get cited. AI tools are pattern-matching on signals of trust that already exist across the web, not rewarding you for effort on your own patch of it.
I ran the same audit for a client, a 14-person accountancy firm in Kent with over 200 blog posts going back nine years, well-written, keyword-solid content. Zero citations across all four tools for any of the 15 questions. Meanwhile a two-person competitor with a scrappy website but three local press mentions and a guest spot on a regional business podcast showed up twice. The content quality gap ran completely the wrong way. That’s the part that stings, and it’s the part most guides avoid because “write more blog posts” is an easier thing to sell than “you need actual press coverage and other people talking about you.”
So what do you do differently
Once you accept that on-page tinkering alone won’t move this needle much, the plan changes shape. Here’s what moved the numbers for my own results between the two audits:
- Pitch yourself for three to five guest contributions a quarter on sites that already have authority, not just your own blog. One decent guest piece did more for my citation rate than twenty posts on my own site.
- Get listed on relevant directories and association pages for your industry, with your name, business name, and specialism spelled the same way every single time. Inconsistent naming across the web confuses these tools.
- Turn your best data or process into something citable, like a calculator, a checklist, or a benchmark. AI tools love quoting a specific number or tool far more than a paragraph of opinion, which is exactly why interactive calculators on your site do double duty as lead magnets and citation bait.
- Build a small number of content pillars around the specific questions your customers ask, rather than scattering posts everywhere, the same approach Joe Pulizzi built his entire content marketing career on, and it holds up just as well for AI visibility as it did for search rankings a decade ago.
- Say yes to podcast interviews and webinars, even small ones. Transcripts get indexed and quoted more often than most blog posts because they’re framed as a direct answer to a direct question, which is exactly the shape these tools are searching for.
Why brand consistency matters more here than anywhere else
One thing the audit made obvious to me: AI tools get confused by businesses that describe themselves differently everywhere. If your LinkedIn says one thing, your website says another, and your Companies House listing says a third, you’re making it harder for a model to confidently connect the dots and cite you as a single trusted entity. Brands that nail this tend to say the same handful of things about themselves everywhere, the way GoPro repeats the same core identity across every channel it touches, or the way Mailchimp keeps its voice and positioning identical whether you’re reading a blog post, a tweet, or a product page. Small businesses rarely think about this kind of consistency because it feels like a big-brand problem. It isn’t anymore.
Go through your own footprint this week: Google Business Profile, LinkedIn company page, any directory listings, your About page. Do they all describe what you do in the same handful of words? If not, that’s an afternoon’s job and it’s a cheap way to make yourself easier for a model to trust.
What a good result looks like
To be clear about what “success” means here, because I think a lot of people picture ranking number one in an AI answer the way they used to picture ranking number one on Google. It’s not that. A good result is being one of two or three sources named, sitting alongside a competitor or a bigger brand, in maybe 20 to 30 percent of the questions someone relevant to your business might ask. Higher than that and you’re probably in a low-competition niche or you’ve got serious existing authority. Lower than 10 percent across a well-built question set and it’s worth treating as a real gap, not a vanity metric.
I’d also say plainly: getting cited doesn’t reliably turn into clicks or leads the way a Google ranking used to. Most AI answers don’t include a link, and even when they do, click-through is low. The value right now is closer to being mentioned in a trusted roundup, brand awareness and trust-building, not a direct traffic channel. Anyone telling you AI citations are the new SEO traffic goldmine is stretching the truth.
If you don’t have time to do this yourself
Running this across a full quarter, tracking it, and turning the findings into an actual PR and content plan takes real hours, and most small business owners I work with don’t have a spare afternoon a week to do it well. That’s exactly the kind of gap where working with an AI consultant for small business pays for itself, because someone who does this audit weekly across dozens of clients spots patterns you’d miss doing it once, on your own, twice a year.
Free resource: The Time Audit Cheat Sheet.
Free resource: The AI Search Visibility Cheat Sheet.
Frequently asked questions
How often should I run an AI citation audit?
Every 8 to 12 weeks with the same set of questions. Results shift as models update and as new content gets indexed, so a single check tells you almost nothing on its own; the pattern across three or four checks is where the useful information sits.
Does having a lot of blog content guarantee I’ll get cited by AI tools?
No. Volume of on-site content matters far less than whether other trusted sites, directories, press, or podcasts mention your business by name. A smaller site with three genuine outside mentions will often out-perform a site with 200 posts and zero outside coverage.
Should I bother trying to rank in AI answers if it doesn’t drive much traffic?
It’s still worth some effort for trust and brand awareness, especially in a buying decision where someone asks an AI tool to compare options before they’ve heard of you. Just don’t treat it as a replacement for the marketing that brings in leads today, like email, referrals, and paid search.
What’s the single fastest fix if my audit comes back with zero citations?
Check whether your business name, specialism, and location are described identically across your website, Google Business Profile, LinkedIn, and any directory listings. Inconsistent naming is the most common and the cheapest thing to fix, and it can be sorted in an afternoon.
Related reading: Why ChatGPT Has Never Heard of Your Small Business and 15 Browser extensions to increase productivity.
Before you change anything, free AI Visibility Checker so you have a baseline to measure against rather than a guess.
For the practical version of this, see write for us about SEO.
