A client of mine — a small dental practice, ranked #3 on Google for “dentist near [their city],” decent traffic, decent reviews — messaged me in a bit of a panic last month. Her daughter had asked ChatGPT to recommend a dentist in the same city for a filling. ChatGPT gave three names. Hers wasn’t one of them. Two of the three weren’t even the biggest practices in town.
That message sat with me for a few days. Because here’s the thing — I’d spent months getting that site ranking well on Google. Fast load times, solid schema, decent backlinks, the whole checklist. And none of it mattered to ChatGPT.
That’s when it hit me: ranking on Google and being recommended by AI are two completely different games, and almost nobody I work with understands the difference yet.
So I Started Testing My Own Client Sites
I didn’t trust one bad result, so I did what I’d tell any client to do — I ran a batch of my own client sites through ChatGPT, Perplexity, and Gemini, using the kind of question a real customer would type. Not “tell me about [business name],” but stuff like “best HVAC company in Dallas that does same-day repairs” or “good WordPress agency for small business sites.”
Out of about a dozen sites I checked, only two showed up consistently across all three AI tools. A few showed up in one and not the others. Most didn’t show up at all, even sites that rank on page one of Google for the exact same search.
A few patterns kept repeating, and they weren’t the things I expected:
- Sites with thin “About Us” pages basically didn’t exist to the AI. If a page doesn’t clearly state what the business does, who it serves, and where, the model has nothing solid to cite.
- Sites with zero FAQ content lost out to competitors that had even a basic FAQ section, even when the competitor’s actual service page was weaker.
- Old, unchanged content got skipped over for a newer, thinner competitor page — recency mattered more than I expected.
- A couple of sites were technically blocking AI crawlers without the owner even knowing it, because of a security plugin’s default settings.
None of that shows up in a normal Google Search Console report. That’s really what pushed me to put something together that could check this stuff automatically, instead of me manually prompting three different AI tools every time a client asks.
Why This Actually Matters Right Now
I’ll be honest, a year or two ago I would’ve filed “optimize for ChatGPT” under hype. Not anymore. People are genuinely asking AI tools for recommendations before they open Google — where to eat, who to call for a plumbing leak, which agency to hire. If your business isn’t part of the answer, you’re not losing a click. You’re losing the conversation entirely, before the customer even sees a list of options.
Traditional SEO still matters, don’t get me wrong. But it’s built around ranking a URL in a list of ten blue links. What AI models do is pick a handful of sources, blend them into one answer, and hand the user a recommendation with no list to scroll through. If you’re not one of the sources, you don’t exist in that conversation.
What I Built to Stop Guessing
After manually checking client sites one by one for a couple of weeks, I built a Free AI Search & GEO Audit Tool so I didn’t have to keep doing it by hand. You put in a URL, and it checks the actual signals AI engines use to decide whether to cite a site — things like schema markup, entity clarity, content structure, and whether your pages are even accessible to AI crawlers in the first place. It hands back a score out of 100 along with the specific gaps behind that score.
It won’t tell you your exact ChatGPT ranking for every possible search — nothing free will, because that changes prompt by prompt. What it does is flag the structural stuff that’s quietly keeping AI engines from citing you at all, which is honestly where most sites are losing before they even get a chance to compete.
How to Actually Use the Score (Not Just Stare at It)
Running the audit takes about a minute. Making sense of the result takes a bit more, so here’s how I walk clients through it:
1. Run the scan first, don’t guess. Paste your homepage or your main service page in. Don’t assume you already know what’s wrong — I’ve been surprised more than once by which page structure actually holds a site back.
2. Read the gap list, not just the score. The number is satisfying, but the real value is the breakdown underneath it. If it flags missing schema or thin entity signals, that’s your starting checklist, not a vague grade.
3. Fix the crawler access issue first, if you have one. This is the one I see most often and it’s the easiest to miss. If a caching or security plugin is quietly blocking AI bots, nothing else you fix will matter, because the content never even gets read.
4. Then go after structure and clarity. Clear headings, a real FAQ section with actual questions people ask, and a page that plainly states what you do and who you serve. AI models cite content they can quote cleanly — vague marketing language doesn’t quote well.
5. Re-run it monthly, not once. AI models retrain and re-crawl on their own schedule. A score from three months ago doesn’t tell you where you stand today.
Mistakes I See People Make With This (Including Ones I Made)
Chasing every keyword instead of answering real questions. GEO isn’t about stuffing “best dentist near me” into your text five times. It’s about writing content that directly and clearly answers the actual question a customer would ask an AI tool.
Ignoring page speed because “Google already ranks me fine.” This one bit me personally. AI crawlers, just like search bots, have to actually fetch and render your page before they can use it. A slow, bloated site on cheap hosting gets skipped or only partially read. When I moved a couple of client sites over to faster hosting and cleaned up their caching setup, their crawl consistency across AI tools noticeably improved within a few weeks. If your hosting is the bottleneck, it’s worth checking — I run most of my own sites on Hostinger specifically because of how fast their WordPress setup loads out of the box, which matters more here than people realize.
Treating this as a one-time project. I made this mistake early on with my own site. I did a big optimization push, saw a bump, then didn’t touch it for two months and watched the visibility quietly slide back down as competitors updated their content and I didn’t.
Assuming a Google #1 ranking means you’re covered. It doesn’t. I’ve now seen enough client examples to say this with confidence — the two systems weigh completely different signals, and a page can dominate one while being invisible to the other.
Final Thoughts
I’m not going to pretend I have this fully figured out — AI search is changing month to month, and honestly, what worked for a client’s site in June looked outdated by August. But the businesses I work with that started paying attention to this early are already showing up in AI answers their competitors aren’t part of yet. That gap is only going to get harder to close the longer you wait.
If you haven’t checked where your own site stands, it’s worth the minute it takes. Run your site through the Free AI Search & GEO Audit Tool and see the actual gaps instead of guessing. And if hosting or site speed turns out to be part of your problem, that’s usually one of the cheaper, faster fixes on the list.
Related reading: if speed or hosting comes up as an issue in your audit, our guide on migrating a WordPress site to Hostinger without losing SEO rankings and our Hostinger LiteSpeed Cache setup walkthrough cover exactly that.
