How You Show Up in ChatGPT and Perplexity Responses

More and more people are no longer using Google—they’re asking questions. Instead of typing in a search query and clicking through ten blue links, they ask ChatGPT, Perplexity, or Google Gemini a question directly and get a ready-made answer, often with a few sources listed below. For businesses, this means that visibility on Google alone is no longer enough, no matter how good their ranking there may be.

A new discipline is emerging, now known as Generative Engine Optimization, or GEO for short. It is still in its infancy, has been studied far less than traditional SEO, and yet is already clearly relevant for any company whose target audience is increasingly turning to AI systems for information rather than traditional search engines.

This post explains how ChatGPT, Perplexity, and similar tools actually source their content, what factors determine whether a source is cited, and what specific changes you can make to your website without having to completely rebuild it. Some of this overlaps with traditional SEO. Other aspects follow their own rules, which are still taking shape.

GEO is SEO for answers, not for clicks

Traditional SEO is designed to get your content as high as possible in search results so that people click on it. GEO is designed to ensure that an AI selects your content as a source for its answer in the first place and mentions you in the process. The difference may sound small, but it isn’t: With SEO, you get a click—and thus a visitor to your site. With GEO, in the best-case scenario, you get a mention without the person asking the question ever visiting your site.

The two disciplines overlap significantly. Technically sound, clearly structured content helps in both cases, but they are not the same thing. A page can rank well on Google and still be completely absent from AI responses if it isn’t formatted in a way that these systems can evaluate. Conversely, a relatively new page without a strong ranking can appear in AI responses if its content provides exactly what a system needs for its answer.

For businesses, this means in practice that the old rule of thumb—“good SEO is enough for everything”—is no longer entirely true. It’s worth considering both disciplines separately, even though many strategies overlap and, in the end, you don’t need two completely separate strategies, but rather an expanded one.

Here's How ChatGPT and Perplexity Access Your Content

ChatGPT uses Bing's index, among other sources, for its web searches, which is why it's worth submitting your sitemap to Bing Webmaster Tools as well, not just to Google. Many companies submit their sitemaps exclusively to Google because that's where most of their traditional traffic comes from, and in doing so, they overlook this second channel.

Perplexity crawls independently and places particular emphasis on timeliness. Both systems, as well as other AI providers, also use their own crawlers to capture content. If you want your own content to appear in AI responses, you should specifically allow the crawlers in your robots.txt file that actually lead to citations, and block only those that collect data for its own sake without benefiting your visibility.

That sounds more technical than it actually is. For most websites, it's enough to check the robots.txt file once and make specific adjustments, rather than blocking all unknown crawlers across the board, as many default security settings do automatically.

Another aspect that is just beginning to gain importance is the llms.txt file, a proposed standard that provides AI systems with a structured overview of a website’s most important content—similar to a traditional sitemap, but designed specifically for language models and formulated in a much more compact way. Not every system supports this file yet, but the trend is clear: The easier it is for an AI system to find the relevant content on a page, the more likely it is to appear in a response later on.

Five Characteristics That Make a Quote Stand Out

The answer first, the reasoning second. When compiling a response, AI systems primarily scan the beginning of a text. An article that answers the central question directly in two to four sentences and then goes into detail is cited significantly more often than one that starts with a long introduction. For many websites, this means reversing the usual writing style, in which the introduction builds suspense rather than providing the answer right away.

Lists and tables. Structured formats are easier for AI systems to extract and incorporate into a response than long, unstructured blocks of text. Where the content allows, a clear list or comparison table is more effective than a continuous paragraph. That doesn’t mean every piece of text has to be turned into a list, but in places where several points are listed side by side anyway, a visible structure is doubly helpful.

Clear, factual language. These systems prefer objective, direct statements over vague or highly opinionated phrasing. The more unambiguous a sentence is, the more likely it is to be included in a generated response. A sentence like “The loading time dropped from 4.2 to 1.8 seconds” is more likely to be included than “The loading time has improved significantly.”

Timeliness. Perplexity, in particular, places a strong emphasis on how recent a piece of content is. An article that is regularly updated with new figures and a visible revision date will be preferred over a static, year-old text, even if the content is nearly identical. Therefore, once you’ve written a good article, you shouldn’t consider it finished, but rather something that needs to be maintained.

Recognizable credibility. A visible author with a brief description, a publication date, and references to sources within the text. These are all signals that these systems use to assess whether content is trustworthy enough to be cited. An anonymous text without a date appears just as unreliable to an AI system as it does to a human reader who doesn’t know whom to trust.

The Underestimated Factor: What Others Write About You

One point that’s often overlooked: AI systems don’t rely solely on your own website. They also draw on platforms like Reddit, G2, or Capterra to verify claims. If you claim a strength on your website that isn’t reflected in reviews and discussions elsewhere on the web, the likelihood that an AI will adopt that claim decreases.

What is said about a company elsewhere on the internet now functions almost like a second application portfolio—one that no one writes themselves, yet still plays a decisive role. This differs from traditional SEO, where a company’s own website was long the most important factor. With GEO, a company’s own website alone is no longer enough, because the systems actively cross-check whether a claim is also confirmed outside the company’s own channels.

For many companies, this represents an unfamiliar shift. Until now, the focus has been almost exclusively on their own website: good copy, solid technology, and compelling arguments. Now there’s an additional layer that can only be partially controlled. Actively managing reviews on platforms like Google, Trustpilot, or industry-specific portals used to be primarily a matter of reputation. Now, it’s also a matter of visibility in the very responses that potential customers see before they even visit a website.

Why this is relevant to your business at all

If you don’t appear on the first page of a traditional Google search, you lose visibility—but at least you still show up somewhere. With an AI-generated answer, there’s hardly any such gradation. Either a company is mentioned as one of a few sources, or it simply doesn’t appear in the answer at all—regardless of how good the content on its own website actually is.

This makes GEO a field with a more “all-or-nothing” nature than traditional SEO. Especially for companies whose target audience makes research-intensive decisions—such as when it comes to services, consulting, or major purchases—the question of whether they appear at all in these AI-generated answers becomes increasingly relevant. Companies that are completely absent here lose not only individual visitors but also their entire role as a source of information in a growing segment of the research process.

A real-world example

We tested this ourselves as part of a client project. For a B2B service company, we first asked ChatGPT and Perplexity the exact same question—one that typical customers would ask before making a purchase decision. The company didn’t appear in either response, even though it ranked on the first page of Google search results for that same topic.

The reason lay precisely in the points mentioned above: The relevant articles began with a long, promotional introduction before the actual question was even answered. There were no visible authors, no update dates, and the few mentions on external platforms sometimes contradicted the statements on the company’s own website. For a system that actively cross-checks whether a statement is confirmed by external sources, this was a clear signal to cite a different source instead. After revising the three most important articles—to include the direct answer first, a visible date, and a brief author attribution—the company appeared in Perplexity responses for the first time within a few weeks. With ChatGPT, it took longer, which aligns with the previously mentioned reliance on the Bing index and demonstrates once again that, despite similar underlying principles, the two systems respond at different speeds.

Measuring is still done by hand at the moment

There aren’t yet any established analysis tools that can show this as reliably as traditional SEO tools do for Google rankings. The most reliable method right now is still to do it manually: Regularly ask the same questions relevant to your business in ChatGPT, Perplexity, Claude, and Gemini, and document which sources are cited. When formulating the questions, do so as a real customer would—not the way you yourself talk about your offerings.

It's time-consuming, but it reveals the actual reality—not just a guess. A simple table is usually sufficient: question, date, system used, sources cited. After a few weeks of repeated queries, a pattern emerges showing which topics are already working and where gaps still exist. This isn't a one-time test, but rather an observation that pays off over time, because the systems themselves are constantly evolving.

A few early tools designed to automate this very process are now coming onto the market. Most of them are still in their infancy, and in some cases their results contradict each other because they use different query methods. We would currently advise against investing a significant amount of your budget in such a tool until the manual method has already provided a clear picture. The ten minutes it takes to perform a manual query currently yield more reliable results than most automated solutions that charge a fee for this service.

These adjustments will suffice for now

You don’t have to completely revamp your existing website to appear in AI-generated responses. A few targeted adjustments are often enough: start important articles with a clear, direct answer; update them regularly instead of writing them once and leaving them as is; use lists and tables where appropriate; and keep a close eye on what’s being said about your company on platforms outside your own website.

It makes the most sense to start with the two or three pages that are most important to your business, rather than overhauling the entire website right away. For these pages, first check whether the main message is actually in the first few sentences, and then add the other points listed above. From there, you can gradually update the rest of the website.

At the same time, it’s worth looking beyond your own website: Which platforms are relevant to your industry—platforms where customers leave reviews or discuss questions? Being actively present there, collecting honest reviews, and correcting obvious misstatements has just as much of an impact on your local search visibility in the long run as your own website does. Together, the two provide a much more complete picture than the website alone.

Conclusion: A new channel, the same basic principles

Visibility in AI responses follows rules that are, in part, different from those of traditional Google ranking, but it’s built on the same fundamentals: clear structure, honest, verifiable statements, and content that actually provides value. If you’re already taking these fundamentals seriously, you’ve already done most of the work.

The rest is a matter of observation and patience. GEO will continue to evolve in the coming years, with new systems, new weightings, and likely better analysis tools as well. Those who start optimizing their content now will have an advantage with each of these changes, rather than waiting to react only after they’ve fallen noticeably behind.

Related: The 5 Areas Where AI Pays Off First in Almost Every Company and GPT-6 Astra: What the New OpenAI Model Means for Your Business · Service: SEO at a Fixed Price

Do you want to know how visible you currently are in AI responses?

We’ll check whether and how your company currently appears in ChatGPT and Perplexity responses, show you where the biggest gaps are, and identify what can be improved in the short term with a reasonable amount of effort. Send us a quick message using the contact form. We’ll respond within 24 hours—personally and without a sales pitch.