SEO for AI: how to show up in ChatGPT and Google answers.

What changes when search becomes an answer

In traditional search, people type a few words and get a list of links. In AI tools, they ask a whole question, sometimes with context, and get a ready-made answer with a few sources cited.

This changes three things for businesses:

• Questions get longer and more specific. Instead of "marketing consulting", people ask "is it worth hiring marketing consulting for a small clinic?".
• There are fewer clicks, and they are worth more. Some people are satisfied with the answer. Those who click through to the source arrive with their question already answered and closer to a decision.
• Being cited becomes a form of recommendation. When the answer mentions your company or your article, it lends you authority.

The name for this work is still settling. Some call it GEO, generative engine optimization, AEO or SEO for AI. In practice, the foundation is the same as SEO: being found, being understood and being seen as trustworthy.

How AI tools choose their sources

When the question calls for current or specific information, the tools run a search on the spot, read the most relevant pages and build the answer from them. That is why being properly indexed is still the first step.

Beyond live search, there is what the model learned during training. Brands that appear often and consistently across the internet tend to be better known by the tools, even when no search is run. This effect is slower and comes from your presence outside the website, the subject of step 3.

The practical summary: if you do not show up well on Google and Bing, you will hardly show up in AI answers. The work starts with the basics.

Step 1: make the website readable for crawlers

AI only cites what it can read. Before thinking about copy, check the foundation:

• Indexed pages. In Search Console, check that the important pages show as indexed. The technical SEO checklist shows the step by step.
• Your website on Bing. Register the website on Bing Webmaster Tools and submit the sitemap. Many people forget, and part of the AI answers depends on that index.
• Text in the HTML. The main content needs to be in the page code, not only inside images, videos or loaded later by scripts.
• Search crawlers allowed. The robots.txt file must not block the crawlers that feed the answers.

These are the crawlers that matter for showing up in answers:

There are also crawlers that collect content to train models, such as OpenAI’s GPTBot. Blocking them is a business decision, and it does not remove the company from ChatGPT search. On Google’s side, the Google-Extended setting controls the use of content to train Gemini and does not affect presence in search.

Blocking a training crawler is a choice. Blocking a search crawler means disappearing from the answers. Check your website’s robots.txt before anything else.

Step 2: write to be cited

AI tools look for passages that answer the question completely and are easy to reuse. The same text that helps a reader in a hurry helps the AI:

• The answer in the first paragraph. Two to four sentences that answer the question in the title, before any context.
• Headings phrased as questions or tasks. "How long it takes to build a website" is easier to find than "Our process".
• Lists and tables. Steps, comparisons and criteria in an organized format are cited more often.
• Numbers, timelines and conditions. "Four to eight weeks" can be cited. "It depends on many factors" cannot.
• One topic per page. A page that tries to answer everything ends up not being the best answer to anything.
• Real frequently asked questions. The questions customers ask on WhatsApp and in meetings are exactly the questions asked to the tools.

An example, on the service page of a company that builds websites:

The second text answers, gives a range and explains the variation. It works for the customer and for any tool that needs to summarize the topic. The same principle applies to the blog: it is the format used in the articles here, with a summary at the top, tables and FAQs.

Step 3: get mentioned outside your website

AI tools cross-check sources. A company that only talks about itself on its own website is less likely to be cited than one that shows up in several trustworthy places. The points that weigh the most:

• A complete Google Business Profile, with the right category, services, photos and replies to reviews. See how in Google Business Profile.
• Detailed reviews. "Fast service for teeth whitening" says more than "great".
• Mentions on industry websites, associations, local press, interviews and supplier lists.
• Content on other channels, such as YouTube videos and LinkedIn articles, with the same company name and description.
• The same details everywhere. Identical name, address, phone number and website on your site, on Google, on social media and in directories.

For businesses that serve a region, this work overlaps with Local SEO. Many questions asked to AI have local intent, such as "best aesthetic clinic near me" or "who builds restaurant websites in my city".

Step 4: make it clear who the company is

To recommend a company, the tool needs to be sure of what it does, for whom and where. Ambiguity lowers the chance of being cited. Four adjustments help:

1. A complete About page. Company name, who is behind it, how long it has been operating, where it serves and which services it offers, in text, without relying on images.
2. An identified author on articles. Name, experience and a link to a professional profile. Content signed by someone who knows the subject is more trustworthy for readers and for search engines.
3. Structured data. Markup such as Organization, LocalBusiness, Article and FAQPage helps crawlers understand the page. It does not guarantee a highlight, but it reduces doubt.
4. The same description everywhere. One sentence that says what the company does, repeated on the website, on the Google profile and on social media. If every place says something different, the AI is left with the most confusing version.

This last point is, at heart, a question of positioning. If the company does not yet have a clear sentence about what it does and for whom, the article on brand positioning helps you get there.

How to measure visits that come from AI

Measurement is still imperfect, but you can already track the trend with what exists:

The monthly test is the most laborious and the most revealing. It shows which competitors appear in your place and which type of page is being cited. That is the map of what to produce next.

Answers vary from one person to another and from one day to the next. So look at the trend over the months, not at a single answer.

What does not work

• Text written for robots. Repeating the keyword and stuffing the page with terms makes it worse to read and does not improve citations.
• Hundreds of AI-generated pages without review. Google treats the mass production of thin content as abuse, and AI answers prefer sources with something of their own to say.
• Expecting miracles from an llms.txt file. It is a proposed standard that some companies have adopted, but the major search engines do not confirm using it. It can exist on the website; it just does not replace the rest.
• Buying mentions and links. Besides the risk of penalties, paid mentions tend to sit on websites nobody reads.
• Treating SEO for AI as a separate project. It is the same SEO foundation, with more care in how you answer.

Frequently asked questions

Conclusion

Showing up in AI answers is the result of three things done consistently: a website that crawlers can read, content that answers directly and a company that is recognized outside its own website. Those who already do SEO well are on the right track; those who do not yet have one more reason to start.

If you want help putting this into practice, take a look at Martin’s SEO consulting.