In this article
For twenty years, "SEO" meant one thing: get a page to rank high enough in Google that a human clicks it. That's still true, but it's no longer the whole job. A growing share of questions now get answered inside ChatGPT, Perplexity, Google's AI Mode, or an AI Overview sitting above the actual results — and those answer engines don't work like a ranking algorithm. They read a handful of pages, decide what's trustworthy and well-structured enough to cite or summarize, and hand the person an answer without a click ever happening. Writing for that is a related skill to classic SEO, but it isn't the identical one.
What actually changed
A traditional search result rewards a page for matching a query closely enough, with enough authority behind it, to rank in the top ten. An AI answer engine isn't ranking ten options against each other — it's picking a much smaller number of sources to actually read, synthesize, and sometimes name, and it's doing that read in real time rather than relying purely on a pre-built index. That shifts what "optimized" means in a few concrete ways: the page needs to answer the actual question clearly and early, not bury it under three paragraphs of preamble; the structure needs to be legible to something parsing it quickly, not just a human skimming it; and the site's broader reputation — mentions, citations, and consistency across other pages, not just backlinks pointing at this one — seems to matter more than it did when a backlink was close to the whole signal.
The numbers back this up. AI Overviews now appear on roughly 48% of tracked Google queries, up from about 30% a year earlier, according to tracking firm BrightEdge. Seer Interactive's analysis of 25.1 million search impressions found organic click-through on queries with an AI Overview fell from 1.76% to 0.61% — a drop of roughly 65%. And ranking well isn't the safety net it used to be: tracking on which pages actually get cited inside AI Overviews shows citation from a top-10 ranking page has fallen from around 76% to somewhere between 17% and 38%, depending on the dataset. Ranking in the top ten used to be close to a guarantee of visibility. It no longer is.
What to actually do differently
Open with the answer. If someone's question is "what does this cost" or "how does this work," say it in the first sentence or two, before the context and the story. That's good writing anyway, but it's load-bearing for AI search specifically, since an answer engine is often skimming for exactly that sentence to quote or paraphrase.
Write real FAQ sections, with real questions. Not seven keyword-stuffed variations of the same phrase — the actual questions a person would type, each with a direct, complete answer underneath. Structured this way (and marked up with the right schema, more on that below), an FAQ section becomes something an AI tool can lift cleanly and attribute, instead of something it has to extract from a wall of prose.
Use structured data, specifically Article and FAQPage schema. This is a layer of markup embedded in the page that explicitly labels what's an article, what's a question, and what's its answer — rather than making a crawler guess from formatting. It doesn't change what a human sees on the page, but it gives both traditional crawlers and AI tools an unambiguous map of the content instead of a guess.
Keep one canonical URL per topic. If the same idea lives at three slightly different URLs (a category page, a tag page, a duplicate), citations and authority split across all three instead of consolidating behind one. A `<link rel="canonical">` tag tells both Google and an AI crawler which version is the real one.
Be specific instead of vague. "Many businesses report improvements" is the kind of sentence an AI summarizer skips over, because there's nothing concrete to cite. A specific number, a named source, or a dated fact is something it can actually quote.
What didn't change
Good writing is still good writing. Being the clearest, most accurate source on a topic is still the best long-term strategy, under either kind of search — AI answer engines are explicitly trying to find and trust accurate, well-structured sources, which is the same bar classic SEO has always claimed to reward and didn't always actually deliver on. The tactics above aren't a replacement for having something genuinely useful to say; they're what makes something genuinely useful actually legible to the systems now standing between a question and an answer.
This isn't theoretical for us — the blog you're reading is built around exactly this: every post gets a canonical link and Article schema automatically, and any post with a real FAQ section gets FAQPage schema generated from it without us hand-writing JSON for each one. If you're curious what the actual tools behind this look like, Surfer SEO is one of the few in our directory built specifically around tracking AI-search visibility, not just classic rankings.
FAQ
What's the difference between SEO and "AI search" optimization?
Classic SEO optimizes a page to rank well against other pages in a results list that a human then clicks through. AI search optimization is about being one of the few sources an answer engine chooses to read, trust, and cite directly — which rewards clear structure, direct answers, and specificity more than it rewards some of the older ranking signals.
Does structured data actually help get cited by AI tools?
It doesn't guarantee a citation, but it removes ambiguity: structured data like Article and FAQPage schema explicitly tells a crawler what a piece of content is and how it's organized, instead of leaving it to infer that from formatting. That makes genuinely good content easier for an AI tool to parse and attribute correctly.
Should I stop optimizing for Google and only focus on AI answer engines?
No — the two overlap far more than they conflict. Clear structure, direct answers, specific facts, and one canonical URL per topic help a page perform better in traditional search results too. Treat AI-search optimization as an extension of good SEO, not a replacement for it.