If ChatGPT isn’t mentioning your business, you’re missing from the data AI models were trained on. AI recommendations run on probability — who appears most frequently across credible sources, not who ranks #1 on Google. Fix it by making your content extractable, your citations cross-platform, and your entity schema clean. Takes 60 to 120 days to move the needle.
Why is ChatGPT ignoring your business?
Your potential customers are typing questions into ChatGPT, Perplexity, and Google’s AI Overviews. They’re getting recommendations — a plumber, an accountant, a digital marketing agency. Just not yours.
The businesses showing up aren’t necessarily the best at what they do. They’re the ones AI models have seen mentioned most consistently, most clearly, across the most credible sources.
This is the new search. And most businesses are completely invisible in it — not because their service is bad, but because they haven’t figured out how this layer works.
Here’s the brutal reality: Google search took 20 years to become table stakes. AI recommendations are taking 2. If you’re not building for this now, you’re already behind.
How do AI models actually decide what to recommend?
Large language models don’t browse the internet in real time when you ask them a question. They generate answers based on statistical patterns in the text they were trained on — billions of web pages, articles, directories, forum posts, and mentions, with a cutoff date.
The key insight: it’s probability, not rank.
When someone asks “who’s the best accountant in Auckland?”, the model completes that sentence based on which accountants appear most frequently and most credibly in its training data. High Google rankings help — because ranked pages get crawled and indexed more often, so they show up more in training data. But the underlying mechanism is pattern frequency, not a direct ranking signal.
This means you can influence AI visibility without necessarily dominating Google SERPs. You need to be mentioned in the right places, the right way, often enough that the model has seen your name in context dozens or hundreds of times.
What four signals build AI visibility?
Four signals consistently separate the businesses being cited from the ones being ignored.
Extractable content — AI models are trained on content they can parse cleanly. Clear headings, direct answers, structured Q&As, numbered lists. If your website reads like a brand brochure — vague value props, stock-photo filler, “we’re passionate about what we do” — the model can’t extract a useful signal from it. Write content that answers specific questions directly. Not “About Our Services” — but “How does [service] work?”, “What does [service] cost?”, “Who is [service] for?”. This is exactly the kind of content architecture we build as part of our SEO service — structured to rank and to be cited.
Cross-platform citations — A single great website is not enough. LLMs see your business name across dozens of surfaces: LinkedIn, Google Business Profile, industry directories, local news, trade publications, Reddit threads, review platforms, podcast transcripts, YouTube descriptions. Each mention is a data point reinforcing that you exist, what you do, and where you do it. Map where you are and where you aren’t. Claim every unclaimed profile. Get listed in every relevant industry directory. Consistent NAP (name, address, phone) across all listings tells the model you’re the same entity everywhere it sees you.
Content freshness — AI models are retrained periodically. The more recently you’ve been mentioned — in new articles, interviews, updated directory listings, published content, refreshed pages — the higher your chance of appearing in newer model weights. Stale entities fade from AI memory. Publish on a real schedule. Update old content with current dates and new information. For tools like Perplexity that retrieve live web results, freshness matters even more — new content can surface within days.
Entity schema — Schema markup tells machines exactly what your business is, who it serves, and what it does. It’s the clearest signal you can send — structured data that any crawler or training pipeline can parse without ambiguity. Implement Organization schema on your homepage, Service schema on each service page, Person schema for key team members, and FAQPage schema wherever you have FAQ content. This is one of the first things we audit in our AI Automation & Consultancy engagements — because most businesses have conflicting entity data scattered across the web.
What does Google actually say about optimising for AI Overviews?
Google’s official guidance is unambiguous: there is no separate optimisation track for AI-generated answers.
From Google Search Central (2024): “Focus on making great content for users. Content that’s helpful, reliable, and people-first.” Google explicitly states that AI Overviews pull from the same index and use the same ranking signals as regular search results.
This is good news. Every solid SEO move — demonstrable expertise, direct answers, E-E-A-T signals, quality backlinks, consistent publishing — also builds AI visibility. They’re not separate disciplines. One investment, two returns.
What tactics are a complete waste of time?
The “AEO” industry has produced a lot of noise. Here’s what not to bother with:
- llms.txt — A proposed standard that no major LLM currently uses for training or answer generation. Add it in 10 minutes if you want to be early, but don’t expect measurable results.
- “AI-specific” schema types — These don’t exist. Anyone selling you AI-specific schema is repackaging standard technical SEO with a rebrand. Use Organisation, Service, FAQPage, and Article correctly.
- “AEO audit” packages — If a vendor can’t explain exactly which AI systems they’re targeting and how they’ll measure the result, walk away. Most AEO is repackaged SEO and content strategy with a scarier acronym and a higher invoice.
- Mentioning your business in chatbot conversations — User conversations don’t affect model weights in real time. LLMs are trained on static datasets with fixed cutoff dates. You cannot chat your way into AI recommendations.
What should you actually do in the next 30 days?
Stop theorising. Here’s the checklist:
Week 1 — Fix your extractability. Read every page on your website. Ask: does this directly answer a customer question? Rewrite the ones that don’t. Add a FAQ section (minimum 3 questions) to your top 3 pages. Implement Organisation and Service schema if you haven’t already — use Google’s Rich Results Test to verify.
Week 2 — Map and expand your citation footprint. Search your business name in Google and count how many independent sources mention you. Claim every unclaimed profile: Google Business, LinkedIn, industry directories. Identify 3 directories where you’re not listed and submit to all 3. Audit NAP consistency across all platforms.
Week 3 — Publish and refresh. Publish one genuinely useful article that answers a question your clients actually ask. Structure it with H2 headings as questions, direct answers, and a FAQ section at the bottom. Update your 3 oldest service pages with current dates and fresher copy.
Week 4 — Measure and brief. Search “[your business name] [your service]” in ChatGPT, Perplexity, and Google. Screenshot what comes up — this is your 90-day baseline. Brief existing clients on leaving specific, service-named reviews. One credible external mention in a trade publication or podcast is worth fifty directory listings. If you want a hand building this system, book a free strategy call — we’ll tell you exactly where your business stands in AI search right now.