AEO vs SEO: what changed, what's new, and what carries over

Reference guide: This post is the take. For the comprehensive reference — what carries over from SEO, what shifted weight, what’s genuinely new, and what no longer moves the needle — see Traditional SEO vs. AI search optimization.

If you’ve worked in SEO for the past two years you’ve watched a new acronym crystallize: AEO. Answer Engine Optimization. Sometimes also GEO (Generative Engine Optimization), AISEO, or LLMO. The vocabulary has converged on AEO, and the question every SEO operator is asking is the same one: is AEO a new discipline I have to learn from scratch, or is it SEO with new branding?

The honest answer is “neither.” AEO isn’t a paradigm shift requiring a rebuild. It’s also not just SEO with a marketing rename. It’s the most demanding evolution SEO has ever had — the underlying work is the same, but the precision requirements are higher, the surfaces are more numerous, and the measurement discipline is genuinely new.

Here’s what’s actually different.

What AEO is

The cleanest definition: AEO is optimizing content for surfaces that return answers rather than links. Traditional SEO optimized for the ten-blue-links SERP — rank higher, get more clicks, drive more traffic. AEO optimizes for surfaces where there are no blue links to click:

• ChatGPT product carousels and conversational responses
• Perplexity citations within synthesized answers
• Google AI Overviews
• Claude and Gemini product recommendations
• Voice assistants reading results aloud

In all of these, the user often gets what they need without ever visiting a website. The optimization shifts from “rank in the list of links” to “be the entity the answer cites or recommends.”

What SEO operators are getting right

Most of what you do for traditional Google SEO carries over to AEO. The fundamentals didn’t move:

• Long-form, well-researched content still wins
• Internal linking still maps topical relationships
• Structured data is more important, not less
• Backlinks still matter (with caveats — see below)
• Page experience and Core Web Vitals still matter
• Technical SEO basics still matter

If you’re running an SEO program that follows the principles in The SEO Playbook circa 2022, about 60% of the work translates directly. You’re not starting over.

What AEO calls out that SEO under-weighted

The 25% of effort that needs reweighting:

Structured data went from “polish” to “load-bearing”

In 2022 SEO, Schema.org markup was a rich-result eligibility item. Adding Product schema with name, price, and availability got you a star rating in the SERP and you moved on.

In 2026 AEO, structured data is the gate. ChatGPT’s product index pulls heavily from Schema.org. Perplexity also reads structured data on the page (though recent testing suggests it doesn’t preferentially weight JSON-LD over surrounding text). A product page without gtin13, brand, category mapped to Google Product Categories, and full offers markup is invisible to roughly half of the AI surface area.

The reweighting: structured data went from ~10% of optimization effort to ~25%. Not new work — just a lot more of it, and applied more systematically.

Google Merchant Center stopped being “for paid Shopping”

This is the unexpected one. Google Merchant Center was the paid-channel infrastructure for the past decade. Then “Surfaces across Google” launched, then ChatGPT integrated Merchant Center data as a primary product source, then Microsoft launched parallel infrastructure for Bing and Edge.

GMC is now organic infrastructure. Catalogs without a validating GMC feed are missing from a significant chunk of AI surfacing — even catalogs that don’t run Google Shopping ads should set it up.

Universal identifiers stopped being “for retailers”

GTINs, MPNs, brand identifiers used to matter for selling into Amazon and the big-box retailers. They’re now decisive for AI agent trust. AI agents treat the presence of gtin13 as a credibility signal — products without it compete with their own resellers and lose, even when the brand owns the canonical product page.

What’s genuinely new in AEO

The 15% of work that doesn’t have a traditional SEO analogue:

Multi-modal content

Multi-modal models (Gemini, GPT-4V, Claude with vision) read product images directly and extract attributes from them. Catalogs with strong contextual imagery — worn-on-model for apparel, scaled-with- context for furniture, in-use shots for tools — outperform what their text-only signal would predict.

This is genuinely new. Google’s image search existed but didn’t parse image content the way multi-modal models do. Every product image is now also a structured data source.

Conversational query intent

Traditional SEO targeted keywords. AEO targets intents expressed as conversational queries: “what’s a good travel stroller for European cobblestones under $400” doesn’t have a clean keyword target. The AI parses the intent and matches against products with those attributes structured in their data.

The new optimization: structure attributes that map to constraint- based queries (durability, weight, size, use case) — not just descriptive keywords.

The “synthesized recommendation” route

Some AI responses recommend products without citing a source URL. “A good wool overcoat in this price range is X by Y” — no link, no attribution path, just the brand surfaced.

The optimization for this route isn’t on-site; it’s brand presence across third-party sources: published reviews, mentions in industry coverage, citations in best-of lists. This is the route most decoupled from traditional SEO. It rewards PR and content marketing more than on-page work.

Multi-surface measurement

Traditional SEO measured Google rankings via Search Console + tools like Ahrefs and Semrush. AEO requires measurement across Google AI Overviews, ChatGPT, Perplexity, Claude, Gemini, Microsoft Shopping, voice assistants, and the long tail of vertical AI surfaces. No single tool covers all of them yet.

The measurement discipline is genuinely new and still being built.

What the AEO label gets wrong

Two pieces of vocabulary the AEO framing introduces that don’t quite match how things actually work:

“Answer engines” implies one type of surface

The AEO framing groups all AI surfaces under “answer engines” as if they all behave the same. They don’t. ChatGPT’s product carousel behaves more like Google Shopping than like Perplexity’s citation flow. Voice assistants behave differently than visual surfaces. Treating “AEO” as a single discipline obscures the surface-specific optimization patterns that actually move the needle.

A more honest framing: each AI surface is its own optimization target, with shared underlying inputs (structured data, content quality, identifiers) and surface-specific weighting.

”Engine” implies algorithmic stability

Traditional SEO treated Google as a stable target — a ranking algorithm that could be reverse-engineered and optimized against. The AEO framing implies AI engines work the same way.

They don’t, yet. AI agent behavior changes faster than Google’s ranking algorithm does. ChatGPT’s product surfacing rules in May 2026 are noticeably different from what they were in November 2025. Optimizing against this quarter’s behavior is an ongoing maintenance task, not a one-time setup.

What to do this week

For an operator coming from traditional SEO:

1. Audit your structured data. This is the highest-leverage item. Every other AI surface optimization compounds on it.
2. Set up Google Merchant Center if you haven’t. Even if you don’t run paid Shopping ads.
3. Add GTIN/MPN to every product. Even private label catalogs need this.
4. Don’t abandon SEO work. The 60% carryover is real. Backlinks, content depth, technical SEO still matter.
5. Add measurement. Pick a tool that tracks AI agent recommendations across surfaces, or build internal tracking.

For the full reference — what carries over from SEO, what shifted weight, what’s new, what no longer moves the needle, and the quarterly workflow — see Traditional SEO vs. AI search optimization.

The reframe to take with you: AEO isn’t the death of SEO. It’s SEO with a higher floor. Operators who already do SEO well need to reweight, learn the new measurement discipline, and add the multi- modal and brand-route work. They don’t need to start over. The ones who don’t already do SEO well will struggle in AEO for the same reasons they struggle in SEO — the underlying data quality work is the same.