Rankings still matter. They're just not the whole scoreboard anymore.

• What is AEO, GEO and SEO?

  Description text AEO (Answer Engine Optimization) is structuring content so AI assistants surface your brand as a direct answer. GEO (Generative Engine Optimization) is making your content the source AI engines cite and synthesize. SEO is traditional search optimization. In 2025 and beyond, all three work as a sequential stack, not separate strategies.goes here

• Is Google still relevant?

  Yes. But it's increasingly stage three of the journey, not stage one.

Here's a number worth sitting with: research from Profound found that only about 6.82% of ChatGPT's top citations overlap with Google's top 10 organic results. Moz ran a separate 2026 study and landed on a similar pattern: 88% of Google's own AI Mode citations don't appear in the organic search results page at all.

Read that twice. The page Google ranks #1 and the page its own AI Mode cites are, almost nine times out of ten, different pages.

If your content strategy treats "rank well" and "get cited by AI" as the same goal with one scoreboard, this is the moment to rebalance how much weight each one gets. They're not measuring the same thing. They're not even built on the same system.

Two different ways an AI can answer you.

When an AI assistant answers a question, it's doing one of two things. Either it's pulling from training data (what it learned during a training run, frozen at a cutoff date), or it's running retrieval augmented generation, RAG for short: searching the live web at the moment you ask, pulling back a set of documents, and generating an answer grounded in what it just retrieved.

Most of the AI search experiences people use daily (ChatGPT with web search on, Perplexity, Google's AI Overviews and AI Mode) run on RAG, not memory. That changes the actual question you should be asking about your content. It's not "is my brand well known enough that a model would have learned about it." It's "would my content get pulled into the retrieval pool for this specific question, right now."

Those are very different problems, and most GEO advice answers the first one while companies are actually losing on the second.

The unit of competition isn't the page anymore.

Classic search ranks pages. A RAG system retrieves passages: specific chunks of text, evaluated and pulled independently of whatever else is on that page. A 2,000 word article with one genuinely strong, self-contained paragraph can get that paragraph retrieved and cited while the rest of the page is functionally invisible to the system.

This is also why keyword optimization is increasingly beside the point. Retrieval systems convert content into vector embeddings, numerical representations of meaning, and match those against the embedded version of the user's question. A page can fail because it never says the exact right "exact match" phrase, while a page that just directly and clearly answers the underlying question wins, even if it never uses the keyword at all.

Practically, that means a page built as a narrative you're meant to read start to finish is harder to retrieve from than a page built as a set of self-contained answers, each one complete enough to be lifted out on its own and still make sense.

Freshness isn't a tiebreaker. It's a filter.

The other piece most teams underweight: how recently something was published or meaningfully updated now functions as a structural gate, not a minor ranking boost. One 2026 analysis found that roughly half of all AI-cited content is under 13 weeks old, and content published in the last 30 days gets cited at roughly 3.2 times the rate of older pages. The same research estimates AI citation has something like a one year half life: a page loses about half its citation potential within about a year of publication, even if nothing about the underlying facts has changed.

Google's own May 2026 guidance on optimizing for generative AI search makes a similar point directly: apply foundational SEO practices to generative AI search, which includes substantive content updates, not just touching a metadata timestamp. A "last updated" date that isn't backed by an actual update is a freshness signal a retrieval system can see through.

Why showing up in other people's content matters.

Retrieval pools draw from across the web, not just your domain. A brand that's mentioned in a Reddit thread, a G2 review, a trade publication, and its own blog shows up in four separate passages a retrieval system might pull from. A brand that only exists on its own site is competing with one entry against a competitor's four.

This is the part of GEO that content teams have the least direct control over and tend to deprioritize for exactly that reason. But it's also the highest leverage move available: earned mentions and third-party citations aren't just a brand awareness play anymore, they're literally how your brand populates more of the index a retrieval system is pulling from.

What this means for how you plan content.

A few concrete shifts follow from all of this:

Write in self-contained, retrievable units. Every section of a page should be able to stand on its own and fully answer one specific question, not depend on the three paragraphs before it for context.

Update, don't just republish. A changed timestamp without a changed fact doesn't move the freshness signal. If a number, a feature, or a claim is more than a few months old, verify it's still true before you rely on it again.

Track AI citations and search rankings as two separate metrics, and budget for both. A page ranking well in Google tells you almost nothing about whether it's being retrieved and cited in AI answers. Don't pull resources off rankings to fund this, fund it as its own line, because optimizing for one doesn't transfer to the other.

Put real effort into earned, third-party mentions. This isn't a PR nice-to-have sitting next to your content plan. It's a direct input into how often your brand shows up across the documents a retrieval system has to choose from.

Rankings Still Matter.

None of this means rankings stopped mattering. SEO is still how a meaningful share of high-intent traffic finds you, and it's still where conversion happens. The shift isn't away from rankings, it's toward treating them as one input among several rather than the proxy for everything. A page that performs well in one system can quietly go invisible in the other, and the fix is giving retrieval its own budget and attention, not taking either one's away.