How AI Is Changing Search — and How to Rank in the Age of Zero-Click

For twenty-odd years, the deal was simple: rank on page one, earn the click, make your case on your own website. That deal is being renegotiated without your signature. Google's AI Overviews now answer many questions directly at the top of the results, and a growing share of discovery happens inside ChatGPT and Perplexity, where there is no page one at all — just an answer, and the handful of sources it names.

This field note is about what that shift actually changes, what it doesn't, and what to do about it in concrete, checkable terms.

What did AI Overviews, ChatGPT, and Perplexity change about being found?

They moved the answer above the links. Google's AI Overviews summarize sources at the top of the results page, while ChatGPT and Perplexity answer conversationally and cite a handful of pages. Discovery increasingly happens inside those answers — so the businesses named in them get the visibility, whether or not anyone clicks through.

Notice what didn't change: these systems don't conjure answers from the void. They retrieve from the crawled web — the same pages classic SEO has always worked on — and assemble a response from the clearest, most credible material they find. The pipeline still starts with a crawlable site and content worth trusting. What changed is the last step: instead of ten blue links auditioning for a click, one synthesized answer names its sources, and everyone else waits in the woods.

Why is being cited the new ranking?

An AI answer names only a few sources, and those citations inherit the trust the answer earns. When an engine cites your page, you are effectively the top result for that question — selected not by keyword match but by being the clearest, most credible passage available. Citability is the new position one.

The mechanical detail that makes this actionable: answer engines retrieve passages, not pages. When someone asks "how much does SEO cost in Seattle," the engine isn't looking for the best overall website — it's looking for the paragraph that most completely answers that exact question. A self-contained block of text that states the question, answers it fully, and carries your name is a passage worth lifting. Forty rambling paragraphs that eventually get around to the point are not.

How do you optimize for answer engines, concretely?

Structure pages so a machine can lift a complete answer with your name attached: question-form headings, a self-contained 40–60-word direct answer under each one, an identical description of your business everywhere it appears, schema markup confirming the facts, and server-rendered HTML — because most AI crawlers do not execute JavaScript.

Here's the working checklist we apply — to client sites and, as you may have noticed from this page's formatting, to our own:

• Turn key headings into the questions customers actually ask, phrased the way they ask them.
• Write a 40–60-word self-contained answer directly under each question — complete enough to quote without any surrounding context.
• Describe your business identically on every page, profile, and directory. Entity consistency is how engines become confident you are who you say you are.
• Add schema markup so machines can verify the basics — who you are, where you operate, what you offer — without inference.
• Server-render your copy. If your words only exist after JavaScript runs, most AI crawlers never see them; they read raw HTML and move on.

That last footprint deserves emphasis because it silently disqualifies otherwise good websites. Crawlers like GPTBot, ClaudeBot, and PerplexityBot generally fetch your HTML and read what's in it — they don't wait around executing scripts the way a browser does. A site that assembles its content client-side can look perfect to a human and nearly blank to the systems deciding whom to cite. Checking for exactly this failure is part of our technical SEO work, and the full answer-engine playbook is what our AI-powered SEO service does month after month.

If nobody clicks, how do you measure success?

Measure what zero-click leaves behind: whether AI engines mention your business when asked the questions you should win, whether branded searches and direct inquiries rise, and whether the leads that do arrive come better qualified. Raw traffic now understates your visibility; citations and qualified leads capture it.

This is a genuine reporting problem, and pretending otherwise is how agencies end up apologizing for flat traffic graphs while the client's phone rings more than ever. A visitor who arrives after an AI answer already explained what you do and what you charge is further along than a curious click ever was. So the scoreboard has to change: ask the engines your customers' questions on a schedule, record who gets named, and watch the trend — alongside branded search volume and lead quality, not instead of them. We do this monthly for clients as the AI Visibility Report, and we run the same experiment on ourselves in public at /tracking, where you can watch a brand-new site try to get cited from zero.

Does llms.txt help you rank?

No. llms.txt is a transparency artifact — a plain-text summary of your site that AI systems can read if they choose — not a ranking factor for Google or any answer engine. Publish one because it's a cheap, honest way to state who you are in a machine-friendly format; distrust anyone selling it as a rankings lever.

We publish one ourselves, for exactly that reason and no other. In a young field, the sales pitches run ahead of the evidence, and llms.txt has become a popular thing to invoice for. It costs an hour and tells the machines the truth about you — worth doing. It does not move rankings or citations, and the honest version of this industry says so out loud.

What should you actually do this quarter?

Pick the ten questions your customers ask most, rebuild the pages that should answer them into question-and-answer form, standardize your business description everywhere it appears, add schema, and confirm your copy is server-rendered. Then test: ask the engines your own questions monthly and note when you start getting named.

None of this replaces the fundamentals — it stacks on top of them, which is the most reassuring thing about the AI shift. The work that made you findable is the same work that makes you quotable; it just needs sharper structure and a new scoreboard. If you want to know where you stand today, the free SEO audit includes an AI visibility check — whether answer engines can see and cite your site at all — and if you'd rather just ask a human, the trackers reply within one business day.