The short version: ChatGPT, Perplexity, Gemini and Copilot are already answering questions about your business, your prices, your opening hours and whether you're "still trading", and most of what they say comes from old data, a scraped directory listing, or a competitor's press release that happened to mention you. You cannot fix this the way you fixed your Google ranking. You have to feed the machines the right answer directly, in the right places, and then wait.
I checked what AI says about my own business. It wasn't flattering.
In January 2026 I sat down and asked five different AI tools the same question: "Who is Lilach Bullock and what does she charge for AI consulting?" I did this on ChatGPT, Perplexity, Gemini, Copilot and Claude, back to back, in one sitting.
Two of them had my day rate roughly right. One quoted a figure I stopped charging in 2022. One told the user I "primarily focus on social media influencer marketing" which was true around 2016 and hasn't been my main work in years. One simply invented a package name I have never offered, a "Silver, Gold, Platinum" tier structure that sounded plausible and was entirely made up, what people in the industry now politely call a hallucination and what I call the AI guessing with confidence.
None of this is a Lilach problem specifically. It is happening to every small business owner reading this, right now, whether you have checked or not. The plumber whose AI-quoted callout fee is from a 2021 Yell listing. The consultant whose old job title from a 2019 conference bio is what gets read out when someone asks an AI "who should I hire for X." The bakery that closed its second branch two years ago and is still being recommended as open by three different chat tools.
Why this is different from an SEO problem, and harder to fix
With Google, you could always do something about a wrong result within days or weeks: update the page, resubmit it, get a backlink, watch the ranking shift. With AI answer engines it does not work that way, and this is the part most people writing about "AI visibility" skip over because it is uncomfortable.
Large language models are trained on a snapshot of the internet from months or sometimes years ago, then topped up with live web retrieval that leans heavily on whatever ranks well, gets cited often, or sits on a high-authority page like Wikipedia, Crunchbase, a national newspaper, or your own LinkedIn profile. Change your website today and there is no guarantee the model's next answer reflects that change next week, next month, or at all before the next training update. You are not talking to a search index that refreshes in real time. You are talking to a system with a memory that lags behind reality, and there is no "resubmit URL" button.
That is the uncomfortable truth nobody selling "AI SEO" packages wants to say out loud: you can do everything right and still watch ChatGPT confidently tell someone the wrong thing about you for another six months.
What you can do about it (in order)
I am not telling you this to make you feel helpless. There is real, specific work here, and I have done it for my own business over the past year while rebuilding my website's income from scratch, so this is not theory.
- Step 1: Run the audit yourself, today. Open ChatGPT, Perplexity and Gemini in three separate tabs. Ask each one: "What does [your business name] do, what does it charge, and is it still trading?" Write down every wrong answer word for word. This takes 15 minutes and most business owners have never done it.
- Step 2: Fix the source, not the symptom. If an AI is quoting an old price, the fix is not a new blog post. It is updating every high-authority page that mentions the price: your Google Business Profile, your LinkedIn company page, Crunchbase if you have a listing, any directory (Clutch, Yell, Trustpilot) that still shows the old figure. These are the pages models trust and re-scrape more often than a random blog.
- Step 3: Give the model one clean, quotable sentence. AI tools love a single, unambiguous, plainly stated fact more than a paragraph of marketing copy. On my own site I now state my day rate as one flat sentence on the services page, not buried in a PDF or a "get in touch for pricing" wall. Ambiguity is what invites a model to guess, and a guess is where hallucinations live.
- Step 4: Get named in places with authority, not just volume. A mention in a respected outlet (a national paper, Forbes, an.edu page, a well-known industry report) carries more weight with these models than fifty low-quality backlinks. If you are trying to get press coverage, my guide to getting a guest post accepted covers exactly how editors think, which is the same instinct that decides whether a mention of you becomes something an AI trusts.
- Step 5: Recheck monthly, not once. This is not a project with an end date. I now run the same five-question check on the first of every month and keep a simple log. It took me three months of corrections before Perplexity stopped quoting my old title, and Gemini still occasionally gets it wrong now.
The bit that moved the needle for me
The single most useful thing I did was not a technical fix. It was giving straight, boring answers in plain language on my own pages instead of clever marketing phrasing. Models struggle with cleverness. "From £X per day for hands-on AI implementation support" gets picked up and repeated accurately. A paragraph about "bespoke transformation journeys tailored to your unique growth story" gets either ignored or garbled, because there is nothing concrete in it for the model to lift.
This is the same lesson I learned the hard way when I published 569 blog posts in 45 days using AI and watched Google index only 5 percent of them. Volume and vagueness do not win with machines. Specificity does. If I had to sum up two years of watching AI tools quote my business back to me, it is this: the models reward the same thing a tired human reader rewards, a clear sentence with a real number in it, and they punish the same thing a tired human reader scrolls past, fluff.
Small things that trip people up more than they expect
A few specific traps I have seen catch other small business owners:
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.
- Old team photos and bios left on an "about" page after someone has left the company. AI tools will happily tell a prospect they will be working with someone who quit eighteen months ago.
- Closed locations that still show as "open" on a directory the model treats as authoritative, even though your own site says otherwise.
- Pricing pages behind a "request a quote" form. If there is no number for the model to read, it will either say nothing useful or, worse, invent one based on similar businesses it has seen.
- Reviews sites where a competitor with a similar name gets your details attributed to them, or vice versa. Worth a search of your exact business name plus "reviews" to see what comes up.
Where this fits with everything else you're doing
None of this replaces good marketing, and it is not a reason to abandon the basics that still work, things like interactive tools that convert visitors on your own site, where you control the message completely and there is no model in between guessing at your prices. Your own website and your own owned channels are still the one place nothing gets garbled by a third party's training data. AI answer checks are a new layer on top, not a replacement for the layers underneath.
If you run a small business and the idea of auditing five AI tools every month sounds like one more job you do not have time for, this is exactly the kind of practical, unglamorous task that a decent AI consultant for small business should be doing for you as standard, not charging extra for as some exotic new service. It is fifteen minutes a month and a checklist, not wizardry.
I am not technical by training, and I have written before about how I use AI every day without being an engineer. This is one more habit I have added to that list, right alongside checking my analytics and my inbox. It takes less time than either.
Frequently asked questions
How do I check what ChatGPT says about my business?
Open ChatGPT, Perplexity and Gemini separately and ask each the same direct question, such as "What does [your business name] do and what does it charge?" Compare the three answers against reality and note every wrong fact. Do this once a month, not once.
Why is AI showing old prices from my website?
AI models are trained on snapshots of the web from months or years back and top that up with live retrieval that favours high-authority pages like Google Business Profile, LinkedIn, Wikipedia and major directories. Updating your own website alone does not guarantee those pages get re-checked quickly, so old pricing can linger for months even after you fix it on your site.
Can I force ChatGPT or Google's AI to correct wrong information about my business?
There is no direct "correct this" button. The realistic fix is updating every high-authority source that mentions the fact, your directory listings, your LinkedIn, your Google Business Profile, and any press or review sites, then waiting for the models to re-crawl and re-train. It usually takes weeks to a few months, not days.
Is this worth the time for a very small business with no marketing team?
Yes, because it costs nothing but fifteen minutes a month and the alternative is a stranger being told the wrong price, the wrong service, or that you have closed when you haven't. For a one-person or small team business, that is a lost customer you never even know you lost.
Related reading: I Fed a Year of My Invoices Into AI. It Told Me Three Clients Were Costing Me Money. and I Audited Every AI Subscription I Pay For. I Cancelled Nine Out Of Sixteen..
