Over the past few months, the question I hear most often is almost always the same: "I ask Gemini about my industry, it lists three of my competitors, and it never mentions me. What should I do?" Behind that question there's also a quiet worry. If a site you spent years building up on Google doesn't appear in the answer an AI writes, was all that effort wasted?
I don't think it was, but on its own it's no longer enough. In this post I'll explain what this thing called GEO actually is, what information AI tools look at when they recommend a brand, and what needs to change on your site. I'll also touch on a few things being sold as "GEO hacks," because some of them really are wasted effort.
The term came out of academia. The paper GEO: Generative Engine Optimization, published by a group of researchers in November 2023, tested how a page could become more visible in the answers of AI-powered search engines. Then agencies adopted the word. In Turkey today, roughly the same service is sold under several labels, such as "GEO," "AI SEO," and the Turkish-language equivalent of "AI SEO."
Don't let the naming confusion wear you out. They're all after the same question: when someone asks an AI something in your field, does your name come up or not?
In practice, I think it's a mistake to treat GEO as a world separate from SEO. Most AI tools search the web while writing an answer and read the pages they find. Those pages are, for the most part, well-written pages that search engines can find. In other words, SEO lays the foundation and GEO is built on top of it.
SEO aims for the top of search results, AEO for the direct answer box, and GEO for a place inside the answer an AI writes. AEO (answer engine optimization) is a term that predates GEO; it was originally used for Google's featured snippets and for voice assistants.
| SEO | AEO | GEO | |
|---|---|---|---|
| Goal | Top position in the results list | Appearing in the direct answer box | Being mentioned in the AI answer |
| Where | Google and Bing results | Featured snippets, voice assistants | Gemini, Perplexity, ChatGPT, AI Overviews |
| How it's measured | Ranking and clicks | Appearing in the box | Being mentioned and cited for the same questions |
| What the user does | Compares and chooses on their own | Reads the answer | Gets a short list and usually picks from it |
In my view, too much time goes into debating the line between AEO and GEO. The work is largely the same. Pages that answer the question clearly and other sites that talk about you carry most of the load in both. Which name it goes by may matter in an agency's proposal. It doesn't change anything on your site.
AI recommends a brand based on two sources: what it learned during training and the pages it finds in the web search it runs at that moment. You can't influence the first from the outside. The second is where you can do the work.
A few places stand out in and around Karşıyaka for children's swimming lessons:
Pay attention to the notes in that answer. The model mentions each brand with a reason, and every one of those reasons is based on information it could find: either a schedule spelled out on the website or a name that comes up in outside sources. An AI can't know which swim school is better. It mentions the ones it can find clear information about.
For answers that use web search, there's also a technical gate. Each tool's own crawler comes to your site and reads the page. OpenAI recommends that sites that want to appear in ChatGPT search allow OAI-SearchBot; Perplexity says the same for PerplexityBot. On Google's side, there's no separate AI bot. According to Google's own documentation, for a page to be a supporting link in AI Overviews, it only needs to be indexed and eligible to show a snippet in regular search; no additional requirements or special optimization are needed (in Turkish).
Make a note of that last sentence. Some of the packages sold as GEO are selling things Google has explicitly said you don't need.
SEO for AI starts with deciding which questions you want to show up for and measuring where you stand today; technical access, content, and outside sources come after that. Whether you work with a GEO agency or do it yourself, the order is roughly this.
The order of a GEO project
Build your question set
Write down 15-30 questions your customers would actually ask an AI. Not "best swimming class," but concrete questions like "swimming class for a 7-year-old in Karşıyaka." Ask each one in several tools and note who gets mentioned.
Open the gate
Check that AI crawlers aren't blocked in your robots.txt file or by your firewall. Make sure your important pages are in Google's and Bing's indexes.
Make it clear who you are
Your business name, address, services, and area of expertise should be written the same way on your site, your Google Business Profile, and your social accounts. Organization and LocalBusiness schema also tells machines this explicitly.
Write pages that answer the question
Put the answer to every important question in two or three sentences at the top of the page, with the details after. Create a new page for any question that has no answer yet.
Get mentioned elsewhere
Industry guides, comparison articles, local directories, the press. Places where the model can verify you.
Ask the same questions again
Repeat your question set in the same tools and compare the results with your starting point. Next month's plan comes out of that comparison.
Step 4 takes the most effort and gets postponed the most. Most service pages open with lines like "quality service, expert staff." An AI can't find anything worth quoting in that sentence. A sentence like "weekdays after 5 p.m. and weekends, ages 5-12, groups of no more than six" contains three separate facts, and all three answer a parent's question.
Be careful with step 5. Buying links or blasting the same text across dozens of directories doesn't work here either, and on top of that it violates Google's rules. A few places where you genuinely deserve to be mentioned are better than a hundred meaningless listings.
The biggest change is in how you write: put the answer at the top of the page, short and concrete. The same information can be written in two different ways, and one of them is far more likely to be quoted.
A page that's hard to quote
A quotable page
You might be thinking, "but my site is for customers, not machines." You're right, and the nice part is that both want the same thing. A parent browsing on their phone also wants to see the answer in the first two sentences. A page written well for AI is usually a better page for people too.
As for schema (JSON-LD), it's a useful thing to set up. But on its own it doesn't make anyone a recommended brand. It tells machines, in an orderly format, the information that's already on the page, and that's it. You can't create information with schema that isn't on the page.
The fastest way to find out is to go through the eight items below; you can check most of them without needing an expert.
AI visibility pre-check
0/8Check off the items; it'll become clear which layer you need to start with.
Almost everyone skips the last item, and I think it's the most important one. One warning, though: if you ask the same tool the same question twice, you may get different answers. Getting excited or discouraged over a single try is misleading. Ask the questions several times, a few days apart, and look at the average.
Some help a little; most are overhyped. A lot of sales talk has built up around GEO, so in this section I'm going to tap the brakes a bit.
llms.txt is a format proposed in September 2024 for presenting a summary of your site to AI tools in an organized file. As the name suggests, it's a proposal. There's no official statement from the major AI companies saying they use this file when choosing sources for their answers. Adding it does no harm. But be skeptical of anyone who says, "We set up llms.txt, so GEO is done."
The same goes for the numbers. The GEO paper reports that some of the methods it tested could increase visibility by up to 40% on its own test set. That figure gets quoted everywhere, but it was measured in a test setup the researchers built, and the paper itself says the effect varies from field to field. You can't work out what will happen in your industry from that number. To be honest, neither can I.
If you're a small, local business, you can do most of the basics yourself; if competition is intense or you have a lot of pages, an agency will save you time. If you do it yourself, the most time goes into rewriting pages and asking the same questions every month and taking notes. On the technical side, there may be a few hours of work for a developer.
If your customers do a lot of research before deciding (healthcare, education, law, B2B), or if you offer many services, working with a GEO agency can speed things up. Even so, no one can guarantee which question you'll be mentioned for, or when; if you see a proposal that promises that, read it twice.
If your customers mostly use ChatGPT, going deep on a single tool is also an option. I explained how ChatGPT chooses its sources separately in my guide to showing up in ChatGPT; our service focused on that side is called ChatGPT SEO. If your site doesn't show up on Google at all, you first need to lay the foundation with an SEO agency and move on to GEO after that.
There's no set timeline. Technical fixes are done within a few weeks, and tools that search the web can read new pages relatively quickly, but no one can say which question you'll be mentioned for, or when. I recommend doing your first serious comparison at the end of the third month.
AEO stands for answer engine optimization. It was first used for Google's featured snippets and voice assistants; today most articles use it to mean the same thing as GEO. The work is largely the same.
It defines the questions your customers will ask, measures your starting point by asking those questions in AI tools, sorts out your site's technical access and content, works on getting you mentioned in outside sources, and asks the same questions again every month. A good agency tells you in writing what it measures and what it can't guarantee.
No. The web searches AI tools run rely heavily on pages that search engines can find. A site that doesn't show up on Google at all will have a hard time showing up in AI answers.
The size of the question set, the number of pages, the technical state of the site, and the scope of the work on outside sources determine the price. That's why a fixed package price quoted without measuring where you stand is usually misleading.
Whichever one your customers use. If you don't know, start by asking the same questions in all of them; seeing which sources each tool shows will point you in the right direction anyway.
Last updated: September 27, 2026. The rules of AI tools change often, and I update this guide as they do.
According to Comscore's September 22, 2026 report, ChatGPT's share of prompts across AI assistants fell from 70% to 50% in six months. Being a source and being named in the answer now need to be measured separately.
Why can two GEO quotes differ several times over? We break down the six line items that make up a quote, from the initial analysis and content creation to external mentions and measurement, and show how to compare quotes side by side.
Using Google's own guide to hiring an agency, we boiled down to eight questions how to spot an agency that promises guarantees, hides its method or sells "Google-approved" tools.
We use cookies to improve the site and measure advertising performance. Analytics and advertising cookies run only if you accept them. Details are in our Privacy Policy and Data Protection Notice.
