A growing share of searches now end with an AI-written answer. Google shows AI Overviews above many results and offers a conversational AI Mode. ChatGPT, Perplexity and others search the web and answer with a summary and a handful of citations.

For website owners this raises two worries. Will people still click through to our site? And how do we become one of the sources these answers cite? This post covers what we know, what we recommend, and what we think is still speculation.

What changes for your traffic

When an answer is complete on the results page, fewer people need to click. Simple factual questions, definitions and quick comparisons are most affected. If a large part of your traffic came from answering those kinds of questions, you have probably already seen fewer visits from search for them.

Searches that require doing something are less affected: booking, buying, getting a quote, comparing specific products, or reading a detailed guide. AI answers often cite sources for these, and the people who click through are typically further along in their decision.

The practical consequence is that raw traffic numbers matter less and the quality of visits matters more. Measure leads, sign-ups and sales from search, not just sessions.

What helps you get cited

The honest summary is that the fundamentals of good SEO still do most of the work. AI search features rely on search indexes and on pages they can crawl, understand and trust.

Be crawlable and fast

If search engines cannot crawl or render a page reliably, AI systems built on top of them cannot use it either. Server-rendered or static HTML, clean internal linking, a sitemap and reasonable performance are still the foundation.

Answer questions clearly

Pages that state a clear answer near the top, then explain it, are easier to quote. Use descriptive headings that match the questions people ask. Put key facts in text, not only in images or PDFs.

Offer something that is not everywhere else

AI answers are built from what many sources agree on. To be cited specifically, a page needs something of its own: original data, first-hand experience, specific examples, clear pricing, real comparisons or expert explanation. Generic content that restates what other pages say gives an AI system no reason to point to you.

Show who is behind the content

Name the authors and their expertise, keep an up-to-date about page, and make your organization's details consistent across your site and the profiles that describe you elsewhere. This helps both traditional search and AI systems decide which sources to trust.

Use structured data where it fits

Schema.org markup for your organization, products, articles, events and FAQs helps machines understand what a page is about. It is not a guarantee of anything, but it removes ambiguity and costs little.

Keep content current

Outdated pages are less likely to be used for answers about current topics. Review your most important pages regularly, update dates only when content actually changes, and remove or redirect pages that are no longer true.

Controlling AI crawlers

AI companies run their own crawlers, and you can control many of them in robots.txt. Several companies separate crawling for model training from crawling for search answers. OpenAI, for example, uses GPTBot for training and OAI-SearchBot for ChatGPT's search features. Google offers a Google-Extended token that controls whether your content is used to train its Gemini models and to ground answers in the Gemini apps, and Google states it does not affect inclusion or ranking in Google Search.

That lets you make a deliberate choice. Many businesses decide to allow search-related crawlers, so they can appear in AI answers, while making their own call about training. Blocking everything can remove you from answers your customers are reading. Check each provider's current documentation before changing your rules, because the names and behaviors keep evolving.

What we would not spend money on yet

llms.txt

llms.txt is a proposed file that summarizes a site for language models, similar in spirit to robots.txt. It is cheap to add, and some documentation sites find it useful. As far as we know, the major AI search products have not committed to using it for ranking or citations, so we would not expect it to change your visibility.

Tricks aimed at AI systems

Hidden text, invented statistics, pages stuffed with question phrases and other tactics aimed at manipulating AI answers have the same future that keyword stuffing had. Search engines have strong incentives to detect them, and they damage trust with human readers in the meantime.

How to measure it

  • In your analytics, look for referral traffic from AI products such as chatgpt.com and perplexity.ai, and track what those visitors do.
  • Search Console includes traffic from Google's AI features within normal web search reporting, so watch impressions and clicks for the pages that matter most to you.
  • Ask your best questions in the major AI tools every few months and note which sources they cite. It is unscientific, but it shows you who you are competing with.

The goal has not really changed. Be the clearest, most trustworthy and most specific source on the topics that matter to your customers. AI search rewards that even more than the old list of ten blue links did.

Want a review of how your site shows up in search and AI answers? Contact us and we will look at the technical basics and your most important pages.