GEO Optimization: How to Get Named in ChatGPT
GEO optimization means making sure AI assistants name your brand in their answer. It works through reachable pages, consistent facts and the sources assistants quote. This guide lists every step with its evidence: what the vendor documents, what a study supports and what is still a guess.
Updated
This page is the how-to. What GEO is and how it differs from SEO is covered in What is GEO?. Here you get the order of work, the evidence behind each step and the cost.
The steps at a glance
Every step carries an evidence grade, the same one used in every Nodpick report: A means the vendor documents it. B means an experiment or a large study of AI answers supports it. C means there is no documented or studied effect yet. Start with A and B.
| Step | Grade | Where the evidence comes from |
|---|---|---|
| Measure first whether you are named | Method | Your own question list, see below |
| Let AI crawlers in | A | OpenAI's crawler documentation |
| State facts clearly and consistently | C | No documented effect, but it avoids contradictory answers |
| Write answer-first pages with numbers and sources | B | GEO study (Aggarwal et al., 2023) |
| Build honest comparison pages | C | Cited pages often contain comparisons; an effect is not proven |
| Get listed on the sources assistants quote | B | AI search source study (Chen et al., 2025) |
| Add an llms.txt | C | No vendor documents using it |
1. Check first whether you are named
Without a baseline you cannot tell after three weeks of work whether anything moved. Write down 10 to 25 questions a buyer asks before choosing a provider, without your brand name. Ask each question in each assistant with web search on, repeat it because answers vary, and record whether and where you are named and who is named instead.
The AI question generator can draft the list. The full routine with a spreadsheet is in how to track AI visibility, and our method explains why every score carries a range.
2. Let AI crawlers in (documented)
OpenAI says OAI-SearchBot surfaces websites in ChatGPT's search features and that sites that opt out are not shown in its search answers. GPTBot collects content for training, and ChatGPT-User fetches a page when someone asks in chat. After a robots.txt change OpenAI says its search systems need about 24 hours to adjust.
# Block training, keep ChatGPT search open
User-agent: GPTBot
Disallow: /
User-agent: OAI-SearchBot
Allow: /Check your firewall and CDN as well as robots.txt, because bot rules there can block crawlers without appearing in the file. The free AI crawler checker reads your robots.txt and shows for each bot whether it gets in.
3. State facts clearly and consistently
Write your category, audience, country and price as plain text on your site and keep them identical everywhere: home page, pricing page and every profile. Contradictions between your sources make it harder for an answer to commit to you. Structured data (Organization, Product) helps when it matches what the page shows, and a Wikidata entry makes you an unambiguous entity (Wikidata in the glossary). Evidence that this alone produces mentions is missing, and Google says you need no special markup for its AI features.
4. Write answer-first pages (studied)
Answer the buyer's question in the first two sentences, then back it up with numbers, prices and named sources. The GEO study by Aggarwal and colleagues measured more visibility in generated answers for exactly this kind of change: statistics, quotations and cited sources. That holds under lab conditions. Whether it holds for your brand only your own before-and-after measurement shows.
5. Build honest comparison pages
Buyers ask for "X alternatives" and "X vs Y", and the sources behind those answers are often comparison pages. A page that says who each product suits, with a price table and a check date, gives an answer something concrete. The effect is not proven, hence grade C. How to build one: comparison pages that AI assistants cite.
6. Get listed on the sources assistants quote (studied)
A large study of AI search answers found that assistants mostly cite third-party sources, not the brand's own site. For questions like "best tools for X" those are roundups, review sites, directories and industry media. See which domains are cited for your questions and make sure you have a correct, current listing or an honest review there. This is the work classic SEO reports do not show.
7. llms.txt: optional and untested
An llms.txt takes half an hour. No vendor documents an assistant using it, and Google says Google Search does not need it. Do it if you have the time, but after steps 1 to 6. Template and details: llms.txt: what it is and how to write one.
What GEO optimization costs
- Do it yourself: it costs time and a spreadsheet. Measuring by hand gets tedious beyond three or four assistants with repeated questions.
- With a tool: Nodpick measures every week. Starter is 49 € net a month (1 brand, 25 questions), Growth 129 € (3 brands, 70 questions), Agency 399 € (15 brands, 220 questions). Every plan asks the same six engines, and billed yearly you pay for 10 months. Details on the pricing page.
- Have it done: website audit 490 €, fix sprint 1,490 €, ongoing care from 990 € / month, all net. You get an offer first (implementation).
How long does it take?
Nobody can promise a date. A robots.txt change takes effect in about 24 hours according to OpenAI; new listings on review sites and in roundups take far longer because they depend on third parties. Plan in weeks and judge the trend, not a single day. Change one thing per week and write it down, or you will not know what worked.
GEO optimization for WordPress, online shops and local businesses
WordPress: many SEO plugins generate robots.txt and schema markup. Look at what they actually output before you add anything, and check the file with the crawler checker.
Online shops: buyers ask about specific products and prices, and marketplaces and comparison articles are often named instead of the shop. More in AI visibility for e-commerce brands.
Local businesses: answers lean on directories, maps and reviews. Consistent name, address and opening hours are the first step, see AI visibility for local businesses. Software companies: AI visibility for SaaS.
Want to know where you stand today? The free check shows whether ChatGPT, Perplexity, Gemini and Claude name your brand and who appears instead.
Common questions
Is GEO optimization the same as SEO?
No, but it builds on it. Crawlable, useful pages still decide which sources get retrieved, and Google says optimizing for generative search is still SEO. What is new is the goal: being named, not just listed. The difference is explained in What is GEO?.
Do I need an agency for GEO optimization?
Not necessarily. You can do steps 1 to 6 yourself. Handing it over pays off when you lack the time or your CMS makes changes hard. Then ask for a fixed scope and price, and for a measurement before and after.
Can I buy recommendations in ChatGPT?
Not as a placement inside the answer. Be skeptical of anyone who guarantees a mention. What you can influence is crawler access, clear facts and your presence on the pages the answers draw on.
How do I measure whether GEO optimization works?
Ask the same buyer questions in each assistant before and after a change, repeat them and compare how often and where you are named. Watch for noise: a few points of movement are normal. The method is in how to track AI visibility.