Clients ask us a version of this question every week: our competitor comes up in ChatGPT and we don't — what's wrong with our site? Usually, nothing. Here's what actually decides the answer, and what we do about it.
Asked in ChatGPT: "Who are the best managed IT providers in Charlotte?"
Several firms come up consistently for Charlotte, including a handful of established local providers with strong client reviews and industry recognition.
Sources the answer was built from
That last chip is the whole story. For broad questions like this one, AI assistants rarely read anyone's website. They summarize what independent sources already say. Your competitor isn't beating you on your own site — they're named in the places the model is reading.
Search used to return ten links and let people choose. AI assistants return one answer and name a few companies inside it. Being on page one still matters, but it no longer guarantees you're in the sentence. And because the assistant is summarizing sources rather than ranking pages, the work that gets you mentioned is partly off your website entirely.
Answers also vary. The same question asked twice, by two people, in two tools, can name different companies. There is no ranking to check and no ad slot to buy. That makes this closer to reputation work than to traditional SEO — which is why we measure it differently, below.
When a client isn't being mentioned, the cause is almost always at the top of this list, not the bottom. We work down it rather than jumping to fixes.
AI crawlers are separate from Google's. A robots.txt rule, a firewall setting, or a bot-blocking service can quietly make a site invisible to them while Google traffic looks fine. Fastest thing to rule out, and more common than people expect.
We run your actual questions, capture which sources each answer pulls from, and check whether you appear in them. Directories, review platforms, roundup articles, trade press, community threads. If you're absent from those, nothing on your website will fix it.
Name, address, services, and specialties should match across your site, Google Business Profile, and third-party listings. Assistants build a picture of who you are from all of it at once; contradictions make you a weaker, riskier thing to name.
This is where structured data and clear question-and-answer content come in. It doesn't put you in the consideration set — it makes you easier to cite accurately once you're there, and it's what wins the narrower, high-intent questions that lead to real inquiries.
Guessing from anecdotes is how these conversations go wrong, so we replace them with a baseline. Together we agree on 15–20 questions a real prospect would actually type — mostly specific and local, not one-word category terms. We record today's answers as a starting point, then re-run the same set monthly and report three things: whether you were mentioned, which sources the answer used, and what changed since last month. That gives us a scoreboard we both trust, and it shows progress on the off-site work before it turns into inquiries.