Mostly no. Most websites are still written for a ranking algorithm, not for an AI system trying to pull a clear answer out of a page. ChatGPT, Google’s AI Overviews, and Perplexity all lean toward pages that give the answer plainly near the top, lay facts out so they’re easy to pick up, and say the same thing everywhere the business appears. None of that is complicated. It’s just that almost no small business site is built that way by default.
How AI search differs from Google ranking
Regular SEO is a competition for position. You’re trying to be one of the ten links on the page, and the searcher picks one. AI search works differently: there’s no list of links, just an answer. Either your information made it into that answer or it didn’t.
This is why a page can sit at the top of Google and still never get mentioned by an AI Overview. The two systems are judging different things.
You will see two acronyms for this, and they are not the same job:
- AEO — answer engine optimization. Getting into Google’s AI Overviews and answer snippets.
- GEO — generative engine optimization. Getting cited by ChatGPT, Gemini and Perplexity when someone asks them a question.
They overlap, but the second one leans much harder on what the rest of the web says about you, not what your own site says. More on that below.
The robots.txt mistake that makes you invisible
This is the one worth checking today, because it is common and it silently costs you everything.
A lot of businesses — or their web people — read a headline about AI companies scraping the web, and block every AI bot in robots.txt. The problem is that “AI bot” is not one thing. Taking OpenAI as the example, their own documentation lists three separate agents doing three separate jobs:
| Crawler | What it actually does |
|---|---|
| OAI-SearchBot | “used to surface websites in search results in ChatGPT’s search features” |
| GPTBot | “crawl content that may be used in training our generative AI foundation models” |
| ChatGPT-User | visits a page “when users ask ChatGPT or a CustomGPT a question” |
Only the middle one is about training. OAI-SearchBot and ChatGPT-User are how you get found and cited. Block all three and you have not protected yourself from much — you have removed your business from ChatGPT’s results.
Anthropic and Perplexity split theirs the same way: ClaudeBot and PerplexityBot for crawling, Claude-User and Perplexity-User for fetching a page to answer someone’s live question. Google separates Google-Extended (AI training) from ordinary Googlebot, so blocking Google-Extended does not affect your normal search rankings.
How to check: open yourdomain.com/robots.txt in a browser. If you see Disallow: / under any of those names, that is a decision someone made — make sure it was a decision and not a copy-paste.
It is a legitimate choice to block training crawlers and allow retrieval ones. What is rarely deliberate is blocking the retrieval ones by accident.
Write the answer first, then the explanation
Most business pages open with a friendly introduction or a mission statement before getting to the point. AI systems usually take the first clear statement of fact they find. If that’s three paragraphs down, or never said plainly at all, there’s nothing for them to take.
The fix is simple: answer the question in the first sentence, then explain it underneath.
Lay facts out so they’re easy to pick up
Lists, tables, and clearly labelled sections are much easier for an AI system to quote correctly than the same information packed into long paragraphs. If a page covers “what’s included,” “how pricing works,” or “who it’s for,” make it an actual list. Not because it looks tidier, though it does — because a list item can be quoted on its own without losing its meaning.
If your site needs JavaScript to show its content, they cannot read it
This is the most expensive technical mistake in this whole area, and it is invisible from the outside.
Googlebot renders JavaScript. It runs a headless Chrome, waits for your scripts, and indexes what appears. The AI crawlers do not. Vercel’s analysis of over 500 million GPTBot requests found no evidence of JavaScript execution at all. GPTBot fetched JavaScript files in roughly 11% of requests and never ran them; ClaudeBot downloaded them in about 24% of requests and never ran them either. As of mid-2026 no AI company has publicly documented adding rendering to its crawler.
The consequence is a site that ranks perfectly well on Google and is a blank page to ChatGPT, Claude and Perplexity. If your site was built as a single-page app — React, Vue or Angular without server-side rendering — this is very likely your situation right now.
How to test it in ten seconds: open your page, right-click, choose “View page source”, and search that raw source for a sentence from your page. If your own words are not in there, the AI crawlers cannot see them either. What you are looking at is what they get.
The fix is server-side rendering or static generation, so the HTML arrives with the words already in it. That is a foundation decision, which is why it usually belongs in a redesign conversation rather than a patch.
Keep your business facts consistent everywhere
AI systems get more confident about a fact when they see it repeated the same way in several places — your site, your Google Business Profile, directories, anywhere else you show up. If your hours or service area say one thing on your website and another on your Google listing, that’s not just confusing for customers. It gives the AI two versions and no way to tell which is right, so it may skip the question entirely.
This sounds trivial. It isn’t. It’s one of the most common problems we find, and it’s usually free to fix.
Being mentioned elsewhere counts
AI models read a lot of the web, not just your website. Being mentioned somewhere else — a local directory, a trade association, a supplier’s site, a review platform — carries real weight. In some cases more than anything you can change on your own page.
This is the part that surprises people: traditional SEO is mostly a game played on your own site, and GEO mostly is not. You can write a perfect page and still lose to a competitor who is mentioned in five places you are not.
Structured data (schema markup) helps too, because it states your business facts in a format built for machines to read rather than leaving them to guess from your page copy. An llms.txt file is worth adding as well, though it’s a newer convention rather than something any AI company has committed to — cheap to do, no guarantees attached.
A quick self-check list
- Does the first sentence of each main page answer the question, instead of warming up to it?
- Have you read your own
robots.txtand confirmed you are not blocking the retrieval crawlers by accident? - Can you find your page’s text in “View page source”, or does it only appear after scripts run?
- Are your services, pricing, and process written as lists where that makes sense?
- Do your business name, address, and hours match exactly on your website and your Google profile?
- Does your site have schema markup describing who you are and what you do?
- Are you mentioned anywhere other than your own website?
This is the newest corner of SEO and it’s still changing month to month. If you want a straight read on where your site stands, that’s what our AI-search optimization service is for — or run the free site check and it will flag the technical half of this list for you.