Learn AI Search & GEO For Free

Generative engine optimisation (GEO) is the process of getting your website/pages named as the source of information pulled from when AI answers questions related to your category/topics.

The output of GEO isn’t a SERP ranking or a featured snippet. It’s your business being named inside an answer that gets generated based on passages pulled from other sites.

SEO and GEO overlap massively. Because yes AI comes from Search, but in GEO we’re optimising to get quoted vs. clicked.

SEO for AI covers the foundational best practices that overlap in both.

SEO → ranking positions
AEO → being featured / snippetted
GEO → being quoted / generated

As shown above it’s relatively nuanced. But the nuances matter.

The majority of your SEO work carries straight over. Structured data. Topic clusters. Page authority. All apply.

But there are nuances within the overlap where what you do matters more to AI systems than it does to users. Which is where understanding AEO comes in.

Our guide to AI covers how to GEO along with the differentiators of AEO.

Stage 1: Retrieval. The AI runs a query and pulls a large pool of documents it might use to answer that query.

Stage 2: Selection. The model reads that pool of documents and decides what to put into the answer it generates.

Your backlinks do their work in stage one. Your content does its work in stage two.

Which is why buying links won’t get you cited on its own. Links get you into the pool. They don’t get you quoted. The page still has to answer the question more directly than everyone else who made it into that pool alongside you.

The model isn’t reading your backlink profile asking “should we quote this page?” It’s asking whether these three paragraphs from Salesforce have the information needed to answer “What is CRM” in a way that satisfies searcher intent.

SEO for AI answers covers all of the information we have above broken into two digestible posts.

Yep. See above.

Links get you into the pool of documents the system retrieves from when it goes looking for an answer. That pool comes out of a search index. And search indexes still rank on links.

But after that they start mattering less directly. Link anchor text matters far less to an AI pulling data than it does to Google.

Does that mean links don’t matter for AI? No.

Backlinks are still overwhelmingly the best indicator of an authoritative source within a topic. That doesn’t change.

Ranking on “CRM definition” is still gonna matter just as much for AI citations. You just have less control over the next step of the process.

Yes. No. Sorta?

Backlinks still matter for AI. They’re what gets you into the pool in the first place.

But mentions are far stronger signals inside of that pool. There is no “follow” attribute on a brand mention. Someone said CRM in the same sentence as Salesforce? That’s training the AI to associate those two pieces of information with each other.

Do yourself a favour and stop thinking about mentions vs. links. They’re not competing for the same signal. Both matter.

SEO for AI mention vs. links dives deep on what we know about how mentions rank for inclusion over links.

Most likely not. Individual links matter far less to being included in an AI generated answer. Your whole backlink profile matters far less.

AI isn’t going to crawl the internet, find your newest guest post on some tax blog, realise with its perfect understanding of language that you rock at SEO and drop your name into the next answer it pulls about CRM.

AI decisions are made in clusters. If seven other sites answered the exact same question more directly, that guest post never gets credited. Period.

SEO for building citations in AI pulls covers exactly why who links to you matters more than how many links you get.

What doesn’t get cited? Anything that makes the model work for the answer.

Long narrative content gets retrieved. Then skipped. Because there’s no extractable answer sitting inside it.

Pages that bury important info in paragraph six rather than frontloading it. Same thing. Skipped.

Our theory is that content AI loves looks a lot like the bite sized answers Google currently tries to surface in featured snippets. Concise answers to clear questions.

Original statistics. Direct definitions. Structured comparisons. Quotes with a name attached. Each one hands the model something it can lift without ambiguity.

SEO for AI: Content types that rank dives deeper on our theory and what types of content get cited most.

Open a tab and start testing.

For now your best bet is actually just going try to get your site to appear in answers across a wide variety of prompts. Keep track of when your site shows up and doesn’t.

When you do start seeing yourself surface try fragmenting your testing. Same question. Hour apart. Different device. Do you appear twice? How often?

The hard truth is measurement around AI is grossly unreliable right now. One query I tested ten different times got me five different answers.

You’re going to want to keep track of trends over specifics. And right now there simply isn’t enough data in a single test to create a trend.

SEO for tracking inclusion in AI answers covers everything we know so far and what we still expect to change.