SEO isn't the whole game anymore: AEO and GEO explained

SEO ranks a page in the blue links. AEO gets a page quoted whole inside an answer box or an AI summary. GEO gets a page cited as a source inside a generated AI reply, on a platform that never shows a ranked list at all. They share a technical foundation but reward different things, and most small business sites are still built for only the first one.

What is answer engine optimization, and how is it different from SEO?

SEO is the twenty-year-old game: rank in Google or Bing's traditional results. Technical performance, backlinks, and keyword-relevant content still drive most traffic for most businesses today, and none of that has stopped mattering.

AEO is narrower. It is the work of getting a specific passage of your content lifted whole into an answer box, whether that box is a featured snippet, an AI Overview, or a voice assistant reading a result out loud. AEO does not care how well the rest of the page ranks. It cares whether one paragraph or one list answers one question cleanly enough to be extracted without editing.

That distinction has gotten more expensive to ignore. Ahrefs analyzed 300,000 keywords, comparing Google Search Console click data from December 2023 against December 2025, and found AI Overviews now correlate with a 58% reduction in click-through rate for top-ranking pages when an AI summary appears, up from a 34.5% reduction the company measured in April 2025 (Ahrefs). Ranking first used to guarantee a click. Now it increasingly guarantees a paraphrase, with your name attached only if your content was the one that got quoted.

What is generative engine optimization, and why does it matter now?

GEO is the newest lane and the one built on the shakiest, least standardized signals. It covers how content gets used and cited inside a generated conversational reply, on a standalone AI chat platform rather than inside a traditional search results page. A page can rank nowhere on Google and still get cited in that kind of answer, if the underlying crawler can read the page and the content answers a specific question with enough detail to be worth quoting.

GEO does not run on backlinks and keyword density the way classic SEO does. It runs on crawl access, on how consistently your business is described across the sources an AI system cross-references, and on whether the specific fact it needs is sitting in plain, readable text rather than buried behind a script that only renders after the page loads. A meaningful and growing share of research and buying-decision traffic now starts in a conversational AI window instead of a search box, and a site with no GEO foundation is invisible in that entire channel regardless of how it ranks anywhere else.

SEO, AEO, and GEO at a glance

None of the three replaces the others. A page built only for SEO can still rank and still never get quoted, because ranking and quoting are scored by different mechanisms that happen to share the same web page as their raw material.

Does FAQ schema still matter, now that Google dropped FAQ rich results?

This is the part of the AEO conversation that changed under everyone's feet in 2026, and most guides on this topic have not caught up.

FAQ rich results stopped appearing in Google Search as of May 7, 2026, ending a phase-out that started in August 2023, when Google first restricted the feature to well-known government and health sites. Search Console reporting for FAQ rich results ends in June 2026, and API support ends in August 2026 (Google Search Central). If your FAQ section was built to earn a dropdown snippet on Google, that specific payoff is gone.

The mistake is concluding the FAQ section itself stopped mattering. It did not. FAQ rich results were a visual feature built on top of structured data markup. AEO and GEO do not read your schema markup to decide what to quote. They read the plain visible text of the question and the answer underneath it, the same way a person scanning the page would. A clearly worded question as a heading, followed by a direct answer in the next two or three sentences, is exactly the shape an AI answer surface is built to lift, whether or not any FAQPage schema sits behind it. Write the section for the reader asking the question. The markup was never the mechanism.

Should you actually publish an llms.txt file?

Probably, but not because it will get read.

llms.txt is a proposed standard: a short, plain-language file at your domain root describing what your business does and linking to your most important pages, meant for an AI system to ingest directly. Adoption of the file itself has grown fast. Originality.ai tracked llms.txt files across more than 3 million monitored websites and found the count rose from 4,088 in June 2025 to 36,120 by May 2026, an 8.8x increase. Actual use of those files by AI systems has not kept pace. Ahrefs pulled server logs across 137,000 domains, found roughly 28% had published a valid llms.txt file, and found 97% of those files received zero requests in May 2026. Of the requests that did arrive, real AI retrieval bots accounted for about 1% of the traffic, while SEO audit tools alone accounted for 21% (Search Engine Journal). Adoption among large companies tells the same story from the other direction: a scan of the full Fortune 500 in March 2026 found only 7.4%, 37 of 500 companies, had implemented an llms.txt file at all (GlobeNewswire).

Put plainly: the file is being published faster than it is being read. It costs under an hour to write and does no harm, so there is no real argument against having one. There is also no evidence yet that it moves whether you get cited. Treat it as cheap housekeeping, not as the deliverable a proposal should be built around.

What actually gets a page cited inside a generated answer?

Three things, none of which is a file format.

Crawl access. Each AI answer engine runs its own crawler with its own user-agent string in your robots.txt, separate from the crawler that builds its live search index. A blanket rule that blocks unfamiliar bots by default can quietly block the exact traffic you are trying to attract. Review each user agent on its own rather than allowing or blocking all of them at once.

Entity consistency. AI systems cross-reference your business name, address, and phone number across your own site, your business profile, and third-party directories before deciding whether to describe you confidently. When those details agree everywhere, the system has an easy time recommending you. When they disagree even slightly, abbreviated one place and spelled out another, it has a cheap reason to cite someone else instead.

Specific, sourced content. A page that states a real number, a real range, or a real named source is more quotable than a page that asserts a vague benefit. This is also just good writing, which is why it repays the effort even for the share of readers who never touch an AI tool at all.

What does this look like for a service business in the Kansas City metro?

A generic "Kansas City" mention on the homepage is not enough coverage for a business that actually serves Overland Park, Olathe, and Lenexa as three separate markets. Dedicated content for each city you serve, LocalBusiness structured data scoped to that specific service area, and NAP details that match exactly across your site and every directory all feed the same underlying signal: this is a real, specific business with a real, specific service area that a search engine or an AI system can recommend with confidence for a hyper-local query. Our own Overland Park page is built on that pattern, and our SEO engagements start with this kind of entity work before touching a single word of new content.

For a deeper walk through the technical side of this, the AI search optimization checklist covers the twelve items we run on every site, including two more things worth skipping and the four numbers that tell you the work is paying off. Recent client work built on the same foundation is in our case studies.

Where do you start if you have touched none of this yet?

Fix the foundation before you touch anything AI-specific. Confirm your pages are indexed with no stray noindex tag, confirm your content renders as real HTML text rather than only after a script runs, and confirm your NAP details agree everywhere they appear. That work alone typically takes a two to four week pass on a site of 20 to 30 pages. Only after that is done does a direct-answer rewrite of your top service pages, or a monthly check of what an AI assistant says about you, add anything measurable. Publishing an llms.txt file and rewriting FAQ schema before the foundation is solid is decorating a house with no plumbing.

Frequently asked questions

What is the difference between AEO and GEO?

AEO is getting one passage of your content quoted whole inside an answer box on a search results page, such as a featured snippet or an AI Overview. GEO is getting cited as a source inside a generated conversational reply on a standalone AI platform that does not show a ranked list at all. AEO still lives inside the search engine you already rank on. GEO lives somewhere else entirely.

Do I need to remove FAQ schema now that Google has dropped FAQ rich results?

No. Removing it does not help you, since the visual snippet is gone regardless of whether the markup stays on the page. The FAQ section itself, written as a clear question followed by a direct answer, is what an AI answer surface actually reads and quotes. Keep the section, keep the schema if it is already there, and stop expecting either one to produce a Google dropdown.

Do I actually need an llms.txt file?

It will not hurt you and it takes under an hour, so there is no reason to skip it. But server log studies covering over 137,000 domains found 97% of published llms.txt files received zero AI requests in a full month, and real AI bots made up roughly 1% of the traffic that did arrive. Treat it as optional housekeeping, not a deliverable worth paying extra for.

Should I block AI crawlers from my website?

Only for a specific reason you can name. Each AI answer engine runs its own crawler with its own user agent, separate from the one that builds its regular search index, so a blanket block can remove you from answers you actually want to appear in. Review each user agent individually rather than allowing or blocking every unfamiliar bot the same way.

How long does it take to start showing up in AI-generated answers?

Plan on roughly two to four months for the first movement after the technical foundation is fixed, since AI systems re-crawl and retrain on their own schedule rather than on request. Technical fixes like indexing and NAP cleanup can be picked up within days, but earning a citation inside a generated answer takes longer because it depends on the source being crawled, trusted, and then selected for a specific question.

Does GEO replace the need for SEO?

No. Most buying-decision research still starts with a traditional search, and the fundamentals, fast pages, real backlinks, and content that matches intent, still drive the majority of that traffic. GEO is an added layer on top of a working SEO foundation, not a replacement for one. A site with no SEO foundation has nothing for a GEO crawler to find in the first place.

Want a plain-English read on where your site stands across all three lanes? Start a project and we will tell you what is actually missing before we suggest anything to fix.