If you’ve looked into AI search lately, you’ve met the alphabet. SEO is search engine optimization. AEO is answer engine optimization, and GEO is generative engine optimization: getting your business into the answers that AI tools write. Agencies now sell all three, often as separate services.
So is this a new job on top of the old one? Mostly, no. But a few things did change, and they matter for a small business. Here’s what stayed the same, what moved and what you can ignore.
Three names, mostly one job
The clearest answer comes from Google itself. Its guide to AI features in Search says “The best practices for SEO continue to be relevant” and that “optimizing for generative AI search is optimizing for the search experience, and thus still SEO” (Google). Its AI Overviews and AI Mode are “rooted in our core Search ranking and quality systems” (same page).
In other words, the pages that do well in search are the raw material for AI answers. The new names describe new places your content can show up. They don’t describe a new kind of work.
What didn’t change
The same pages, read the same way. Engines and AI tools still need to find your pages through links, be allowed in, read the content and understand it. We walk through those steps one by one in Agent readiness: where to start.
No special files or code. Google says plainly: “You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search” (Google). If someone tells you AI search needs a new file format before anything else, ask them where the engine says so.
Quality still decides, and tricks still backfire. Google’s spam policies now name “attempting to manipulate generative AI responses in Google Search” (Google). Bing says content that tries to manipulate the AI behind Bing and Copilot “may result in reduced visibility or removal from search experiences” (Bing).
What did change
Fewer clicks per search. When Google shows an AI summary, people click a regular result in 8% of searches, against 15% without one, and only 1% click a link inside the summary (Pew Research Center). Showing up is no longer the same as being visited. AI answers do show sources: Google says its AI features include “prominent, clickable links to relevant web pages that support the information in the response” (Google). Being one of those sources is the new goal.
More readers, with separate switches. It used to be mostly Googlebot and bingbot. Now each AI company runs separate bots for training, for search and for fetching pages on request, and blocking one doesn’t block the others. We sort them out in robots.txt in 2026. Google also added a setting in Search Console that keeps your content out of its AI features. Google says it “isn’t used as a ranking or inclusion signal affecting other parts of Search” and “doesn’t affect AI training” (Google). It’s a real choice now, with a real cost: opting out means you’re not among the sources in those answers.
Many new readers don’t run JavaScript. Google renders the scripts on your pages. Many AI readers don’t: “none of the major AI crawlers currently render JavaScript”, Vercel found when it measured them (Vercel). A page that only shows its price after a script runs was fine for Google. For the rest, it’s empty. Putting the facts in the HTML your server sends went from good practice to necessary.
Answers are turning into actions. The biggest change isn’t in search at all. Assistants increasingly go on to do the next step for the person: compare, book, buy. In July 2026, 42% of US consumers told Adobe they had used an AI assistant to shop online, and shoppers an AI sent converted 60% better than other traffic (Adobe). AEO and GEO are about being quoted. Agent readiness goes one step further: can an agent act on what it found? We look at what agents take in from a page in What an AI agent reads when it lands on your product page.
What you can safely ignore
Numbers for where you stand in AI answers. These tools can only sample a few questions from the outside. Google’s guide is blunt: “No third-party tool has access to our internal ranking or AI systems”, and it advises you to “Be wary of third-party tools that promise ranking success” (Google).
Schema markup as a way into AI answers. Structured data still earns rich results and merchant listings in Google. But Google says “there’s no special schema.org markup you need to add” for its AI features (same page). Add markup for what it does in search, not as a key to AI.
AI text files. We don’t recommend an llms.txt file for visibility. Google says its Search ignores them, and no other engine says it reads one (same page).
Hidden instructions for AI. Text meant only for AI readers, hidden from people, is exactly what the spam policies above describe. It risks your search visibility for no documented gain.
So what do you do differently?
Less than the new names suggest. Keep doing the basics, and do them more strictly than before: links to every page that matters, the right bots let in on purpose, the facts in the HTML and visible on the page. That covers search and AI answers at once.
Then add the one thing that is actually new: making sure an agent that arrives on your site can do what the customer sent it to do. That’s agent readiness, and it builds on everything above rather than replacing it.
