Getting Your Business Seen in ChatGPT and Google AI Overviews: An Intro to GEO

The way customers find you is changing. They used to type 'dental clinic in Beşiktaş' and click one of ten links. Now they ask ChatGPT the same question or read the AI summary Google shows at the top, and often leave with a handful of business names. The work of getting into that handful is called GEO.

The good news: most of GEO is the natural result of well-done SEO and an honest, clear website. The less good news: some approaches in this area promise sure-fire formulas, and part of it is still uncertain. Nobody can guarantee a place in AI answers. Below, we separate what's known from what's observation or educated guesswork.

What is GEO, and how is it different from SEO?

GEO stands for Generative Engine Optimization. The aim is to be shown as a source, or to have your name mentioned, in answers generated by assistants such as ChatGPT, Gemini, Perplexity and Claude, and by Google's AI Overviews and AI Mode.

In SEO, the goal is to rank high and get clicked. In GEO, the goal is to be inside the answer. The difference looks small but matters: an AI answer blends several sources into one text, and some users read it and decide without clicking any site. So sometimes you can be 'chosen' without getting a visit.

The link between the two is stronger than people assume. Most AI search tools consult a search index in the background and read the pages they find when generating an answer. Google's AI Overviews rely on Google's own index. So a page that never appears on Google is unlikely to show up regularly in AI answers. GEO doesn't replace SEO; it builds on it.

How do AI assistants choose sources?

Honesty matters here: companies don't explain their source selection in detail. What we have is the general framing of official statements and trends observed in practice.

First, separate two mechanisms. One is what the model learned during training. If a model saw enough consistent information about your business in its training data, it may know you without searching. This layer changes slowly and is hard to influence directly.

The other is live search. The assistant takes the question, searches the web, reads a few pages and composes the answer from them. This is largely the layer you can influence.

In live search, pages that stand out often share these traits:

• They contain a passage that answers the question directly and briefly.
• Their information doesn't contradict other sources; the business name, address and what it does are the same everywhere.
• They look current; dates, prices and scope aren't outdated.
• They give concrete information: numbers, durations, areas, conditions.
• They're technically readable; bots can access the page and see the text.

For local questions ('best hairdresser near me'), the picture is a little different. Google Business Profile, map data and customer reviews appear to play an important role. For a local business, effective GEO work often starts with a complete business profile and genuine customer reviews.

Forums, industry directories, news sites and comparison articles are also frequently cited sources. What others write about you can matter as much as what you write on your own site.

Writing content that can be quoted

AI often reads a page in pieces rather than as a whole, and pulls the most suitable piece into its answer. So every section needs to make sense on its own.

One effective habit is to make the question the heading and give the answer in the first sentence. Under a heading like 'How long does a boiler service take?', the first sentence should state your real timing, for example 'A standard boiler service with us takes about 45 minutes to an hour.' Explanation, exceptions and detail come afterwards. A paragraph opening with 'The answer depends on many factors' is a weak candidate for quoting.

Concrete information can raise the chance of being quoted. 'Affordable, fast service' answers no question. A line like 'Same-day service in Kadıköy, Üsküdar and Ataşehir; servicing costs 1,500 to 2,500 TL in 2026' answers several. These must, of course, be your real and current figures.

Passages shouldn't lean on each other. A paragraph that starts 'As mentioned above' or 'this method' loses its meaning when read alone. Name what you're talking about again in each section.

Some content types work especially well:

• Clear definitions: a direct two or three sentence answer to 'What is X?'
• Step-by-step processes: application, setup, delivery stages.
• Comparisons and limits: stating who your service isn't right for is a trust signal for a system looking for balanced sources.
• FAQ sections built from real questions.
• Original information: details from your own process and field observations that can't be found elsewhere.

That last point may be the most important. If you write something that's been written a hundred times online for the hundred-and-first time, there's no reason for AI to pick you. Content that answers your own customers' questions from your own experience stands out.

Showing publication and update dates on your pages also makes freshness clear.

Entity clarity: consistent brand information

AI systems treat your business as an 'entity': a record with a name, a location, a field of activity and relationships. They can compare information about you across multiple sources. If it's consistent, they identify you with more confidence; if it's contradictory, they either confuse you with someone else or leave you out.

What to check:

• The business name spelled the same way everywhere. If your registered name is long, decide whether you'll appear under the short name or the full one, and stick with it.
• Address, phone number and website identical on your site, Google Business Profile, social media accounts and industry directories.
• A one-sentence description. For example: 'Kodyva is a web agency offering ready-to-brand websites and custom-built websites.' This sentence should appear consistently on the homepage, the About page and in social media bios.
• An About page that says who you are: founders, founding year, location, areas served.
• A presence on external sources: professional chamber listings, industry directories, local news, customer reviews.

If your name resembles another business, always include distinguishing information (city, sector) in your description. Details of similarly named companies can get mixed up in AI answers.

Structured data and technical accessibility

Structured data presents your business information in a form machines can read without hesitation. Organization or LocalBusiness schema, plus a sameAs field listing your social profiles, contributes to entity clarity. We covered this with examples in a separate article.

On the technical side, the most overlooked issue is bot access. AI companies' crawlers generally follow the rules in your robots.txt file, and if some are blocked your content can't be read. A few names to know (check the current list in each company's documentation):

• OAI-SearchBot and ChatGPT-User: ChatGPT's search and on-behalf-of-user page reading
• GPTBot: the crawler OpenAI uses for model training
• PerplexityBot and ClaudeBot: the crawlers of Perplexity and Anthropic
• Google-Extended: a token that controls whether your content is used for Gemini models. Google's AI Overviews, by contrast, run on the regular Googlebot.

Blocking training bots is a matter of preference. Blocking search bots can directly reduce your visibility. Some hosting and CDN services, such as Cloudflare, offer settings that can block AI bots by default. If your site sits behind such a service, check the settings once.

Then there's JavaScript. On sites where content is loaded later in the browser, many AI crawlers are reported not to execute JavaScript and therefore can't see the text. Having the page text prepared on the server and delivered in the HTML is the safe route.

A proposed standard called llms.txt also comes up often here. Its impact is still unclear; we discussed what it is and whether it's worth adding in a separate article.

How do you track visibility?

Measuring AI visibility is harder than measuring SEO. There's no Search Console for ChatGPT. But you can build a rough picture.

The first method is manual testing. Prepare a list of 10 to 20 questions your customers might ask, such as 'Can you recommend an accounting firm in Ataşehir?' or 'How much does a website cost for a small business?' Ask them once a month in ChatGPT, Gemini, Perplexity and Google's AI Mode, and note in a table whether your name appears and which sources are shown. Answers vary by person and over time, so look at the trend across months rather than any single result.

The second is traffic sources. In Google Analytics, watch referring sites such as chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. Some clicks from ChatGPT also add a utm_source=chatgpt.com parameter to the URL.

The third is Search Console. Impressions in Google's AI Overviews are counted within the overall search performance report, but not shown separately. Still, if impressions rise while clicks fall for certain queries, that can be a hint. We looked at how AI Overviews affect traffic in a separate article.

Don't forget the simplest method: add an 'AI assistant (ChatGPT etc.)' option to the 'How did you hear about us?' question on your contact form. The customer's own answer is information no tool can give you.

8 steps you can take this week

1. Write a one-sentence description of your business and make it identical on the homepage, About page, Google Business Profile and social media bios.

2. Compare your name, address and phone number across your site, Google, social media and any directories you're listed in. Fix any that differ.

3. Collect the 10 questions customers ask you most. For each, write an FAQ entry or page section whose first sentence is the answer.

4. Add concrete, accurate information to your service pages: a price range or the factors that set the price, timing, scope, areas served.

5. Check your robots.txt file and, if you have one, your CDN settings. Are AI search bots blocked without you knowing?

6. Check or add your Organization or LocalBusiness schema and validate it with the Rich Results Test.

7. Complete your Google Business Profile and reply to recent reviews.

8. Create your 15-question test list, record the first measurement today and start watching AI sources in Analytics.

None of these steps is a trick. As AI tools change, some of today's tactics will age. A site that gives clear, consistent, verifiable information will keep working regardless of which tool picks which source.