What is generative engine optimization (GEO)?
How AI assistants choose which sources to cite, and what you can change.
On this page
Short answer
AI SEO is a label people use for two separate jobs. The first is using AI software to get search work done faster: grouping keywords, drafting titles, sorting crawl data. The second is preparing a website for search engines that reply with AI, such as Google’s AI Overviews and AI Mode, ChatGPT search, Perplexity and Gemini, where the buyer reads a written answer with a few sources attached. For a business owner the second job usually has more at stake, because it decides whether your company appears in that written answer at all.
The two meanings compared
| AI as a helper for SEO work | SEO for search that answers with AI | |
|---|---|---|
| The question behind it | How can our team do SEO faster? | How do we get named and linked in AI replies? |
| Who sees the result | your own team | buyers asking Google, ChatGPT or Perplexity |
| Typical tasks | keyword grouping, first drafts, meta tags, crawl and log summaries | pages that answer up front, consistent company facts, mentions on other sites, crawler access |
| What success looks like | hours saved, more pages shipped without errors | how often you are named and linked across a fixed question list |
| What usually goes wrong | thin or wrong pages published in bulk | judging progress from one reply |
When an agency or a tool vendor offers “AI SEO”, ask which column they mean. The work, the price and the way to check results are different.
What changes for a business website
Google’s AI features build on regular search, and its own documentation says so. The page on AI features and your website states that there are no additional requirements to appear in AI Overviews or AI Mode, and that a page has to be indexed and eligible to be shown with a snippet. So the old groundwork still counts: a site Google can crawl and pages that match what people search.
What changes is the shape of the result. A buyer no longer scans ten links. They read a short reply built from several pages and maybe click one source. That has four practical effects:
- Each page needs a quotable core
A reply lifts a sentence or a list from a page. A page that answers its own question in the opening lines gives it more to lift than one that starts with company history.
Read more
Our AEO guide covers the page layout.
- Other websites speak for you
Replies are often assembled from pages that are not yours, such as list articles and review sites.
Read more
If your company is absent from them, a competitor may be named instead. The GEO guide lists the off-site work.
- Crawler access matters in a new way
OpenAI’s documentation on its crawlers describes OAI-SearchBot as the crawler for ChatGPT search.
Read more
A site that opts out of it is not shown in ChatGPT search answers, though it can still appear as a plain navigational link. GPTBot is a separate crawler tied to model training. A robots.txt rule copied from a forum can drop you from one product without you noticing.
- Reporting gets murkier
Google counts traffic from AI Overviews and AI Mode in the Search Console Performance report under the Web search type, together with ordinary results.
Read more
To see what AI replies say about you, you need your own record of questions and answers.
Where AI tools help in SEO, and where they do not
Using AI in the work is allowed. In a February 2023 post Google wrote that appropriate use of AI or automation is not against its guidelines. The line it draws is about purpose: content generated mainly to manipulate rankings breaks its spam policies.
Tasks where a language model saves real time:
- sorting a long keyword export into topics before a person checks the groups;
- suggesting several title and description options for a page, which an editor then picks and corrects;
- turning a crawl export into a short list of broken links, missing tags and duplicate pages;
- drafting schema.org markup from facts you supply;
- comparing your company facts across pages to find mismatched prices, years or addresses.
Tasks where it lets you down:
- Facts about your business. Google’s guidance on generative AI content reminds site owners that these models predict a likely sequence of words and do not retrieve facts. It asks for a manual fact check before publishing, and that includes titles, meta descriptions, structured data and alt text.
- Pages in bulk. Google’s spam policies name, as an example of scaled content abuse, using generative AI tools to produce many pages that add no value for users.
- Search demand. A chatbot will happily invent a monthly search volume. Take volumes from a keyword planner or from your own Search Console data.
- Judging your own visibility. Asking a chatbot whether your site is “optimized for AI” gets you an opinion. It is no substitute for a measurement.
When an agency or a tool vendor offers “AI SEO”, ask which column they mean.
How to check your visibility in AI answers
You can run a basic check in an afternoon with a spreadsheet:
- Collect 20 questions buyers type before they choose a supplier like you: “best [service] in [city]”, “[service] cost”, “[your company] vs [competitor]”, plus a few problem questions with no brand in them.
- Pick the assistants that matter in your market. For most service businesses that means Google, ChatGPT with search on, Perplexity and Gemini.
- Put each question to each assistant three times, from the country your buyers are in. Wording shifts from run to run, so a single answer is a sample, not a verdict.
- For every run, record five things: whether you were named, whether you were linked, which competitors were named, which pages were cited, and whether the facts about you were correct.
- Keep the list fixed and repeat it once a month. A growing count is progress. A wrong price or a missing service tells you what to fix first.
Our guide to AI citations explains how each assistant displays its sources, which helps when you fill in the fourth column.
Common mistakes
- Treating AI SEO as a way to publish more. Hundreds of generated pages with nothing new in them risk Google’s scaled content abuse policy and give buyers little.
- Publishing drafts without checking numbers. A wrong price in a page can end up quoted in an AI reply.
- Dropping the basics. Google says SEO fundamentals stay relevant for its AI features. A page that is not indexed cannot be a source.
- Blocking the wrong crawler. Check robots.txt and any firewall rules for the search crawlers of the products you want to appear in.
- Trusting a single score. Read the actual replies and the pages behind them before you act on a dashboard number.
FAQ
Is AI SEO the same as GEO?
They overlap. In its second meaning, AI SEO covers roughly what generative engine optimization describes: being named and cited in AI replies. LLM SEO narrows the focus to what language models know and say about your company. The AI search glossary defines each of these terms in a line or two.
Will Google penalize text written with AI?
Not for the tool alone. Google judges the content and the purpose. Text produced at scale mainly to rank, with little value for readers, falls under its spam policies whoever or whatever wrote it.
Do I need special software for AI SEO?
No. A spreadsheet and a few hours a month cover the checks above. Paid trackers help once you follow many questions in many assistants, but read the replies yourself before trusting their charts.
Can an agency do this for us?
Yes. Our AI search optimization service covers the monthly measurement, page fixes and work on outside mentions, and ends each month with a report.
