What is Answer Engine Optimization (AEO)?

I asked ChatGPT to recommend a spicy hot sauce with ghost pepper and smoky notes for grilled meats. It recommended Heatonist. By name. With a description of their products, flavor profiles, and a link.

Then I asked it to recommend products from about 50 other online stores I'd been testing. Most of them? Nothing. ChatGPT had no idea they existed.

That gap -- between stores AI can talk about and stores AI has never heard of -- is what Answer Engine Optimization addresses.

The short version

AEO is the practice of making your store's content readable, structured, and citable by AI assistants. ChatGPT, Perplexity, Google's AI Overviews, Copilot -- these tools don't return a list of ten blue links. They answer questions. They cite sources. They recommend products.

If your store's content is structured in a way these systems can parse, you get mentioned. If it's not, you don't.

That's the whole idea. No mystery to it.

This isn't theoretical anymore

ChatGPT has over 800 million users. Perplexity's traffic is growing roughly 3x year over year. According to Amsive's 2025 data, AI-driven retail traffic sits at about 4.87% of total visits. Search Engine Land reported it's already hitting 18% in certain product categories.

Those numbers are still small compared to Google organic, but they're growing fast. And the stores getting that traffic right now are the ones that set themselves up early.

Here's what I think most merchants miss: this traffic converts differently. Someone asking ChatGPT "what's the best hot sauce for grilled meats" is further down the purchase funnel than someone typing "hot sauce" into Google. They're describing what they want. They're asking for a recommendation. That's buyer intent at its clearest.

How AEO differs from traditional SEO

SEO is about ranking in a list. You optimize a page, target a keyword, build backlinks, and try to show up on page one. The user sees ten results and picks one.

AEO is about getting cited in a conversation. There's no list. The AI picks for the user. It synthesizes information from multiple sources and delivers a direct answer, sometimes with a recommendation and a link, sometimes without.

The implication is blunt: in a traditional search, being result number three still gets you clicks. In an AI answer, being the one store that gets cited gets you everything. Being the store that doesn't get cited gets you nothing.

There's no page two of a ChatGPT response.

What makes a store visible to AI

I've run the SEOMelon diagnostic on over 50 stores at this point. The ones that score well share a few things in common.

Structured data that actually says something. Product JSON-LD with complete fields -- name, description, price, availability, brand, reviews, images. Not just the bare minimum your platform auto-generates, but rich, complete schema that gives an AI system real information to work with.

FAQ content on product and collection pages. When someone asks an AI "what's the best moisturizer for sensitive skin," the AI looks for pages that directly answer that kind of question. FAQ schema with real answers to real buyer questions is one of the strongest signals you can send.

Product descriptions that answer questions instead of just describing features. There's a difference between "Our hot sauce is made with ghost peppers" and "If you're looking for a smoky, high-heat hot sauce for grilling, this ghost pepper blend delivers 150,000 Scoville units with a deep mesquite finish." The second one matches how people actually ask AI for recommendations.

Consistent, complete information across your entire catalog. AI systems build understanding of your brand across many pages, not just one. If half your products have thin descriptions and no schema, the AI's overall confidence in your store drops.

What makes a store invisible

The flip side is straightforward. Thin product descriptions -- one or two sentences copied from a supplier. No FAQ schema anywhere on the site. Missing or incomplete JSON-LD. No content that matches the natural language questions people ask AI assistants.

Most stores I've tested fall into this category. I ran diagnostics on Commondeer. Score: 0. ForestWholeFoods. Score: 0. These aren't bad stores. They're just not structured for this new channel.

The diagnostic I built asks questions the way a real buyer would. Things like "Can you recommend some steamy romance ebooks with a forbidden love trope?" or "I want a really spicy hot sauce with ghost pepper and smoky notes for my grilled meats." Then it checks whether an AI assistant would mention your store in its response.

Most stores score between 0 and 50 out of 100. A few score 100. Heatonist is one of them -- ChatGPT actively recommends their products by name. The difference isn't luck. It's structure.

What separates a 0 from a 100

I looked at what the high scorers have that the low scorers don't. It comes down to three things.

First, depth of product content. Heatonist doesn't have a two-sentence description on their product pages. They have detailed tasting notes, heat levels, ingredient stories, pairing suggestions. That's the kind of content an AI can extract and synthesize into a recommendation.

Second, FAQ content with real buyer questions. Not "What is your return policy?" but "What hot sauce should I use for tacos?" and "Which sauce has the most heat?" -- questions that match how people actually talk to AI assistants.

Third, complete structured data. Their JSON-LD isn't just the default Shopify output. It includes reviews, ratings, detailed product attributes. When an AI crawls their site, it gets a complete, machine-readable picture of what they sell and why someone would buy it.

The stores scoring 0 are typically missing all three. They're relying on whatever their platform generates by default, which usually isn't enough.

Where to start

If you're wondering where your store stands, I built a free diagnostic tool. Go to seomelon.com/llm-check, enter your URL, and it'll run real buyer-intent questions against AI models to see whether your store gets recommended.

It takes about 75 seconds. You'll get a score out of 100 and a breakdown of what the AI knows (or doesn't know) about your products.

No signup required. No sales pitch on the other end. I built it because I wanted merchants to see the problem before I tried to explain the solution.

If you score low -- and most stores do -- that's actually fine. It means there's a clear opportunity to get ahead of competitors who haven't figured this out yet. AEO is early enough that improving your structured data and product content now puts you in a small minority of stores that AI systems can actually work with.

The stores that move first on this will have a compounding advantage. AI systems learn and update. Once they know your brand and your products, that knowledge persists. The longer you're visible, the more entrenched your position becomes.

SEO took stores years to figure out. AEO is simpler in some ways -- the signals are more concrete, the fixes are more mechanical. But the window for being early is closing as more merchants and more tools catch on.

The question isn't whether AI shopping assistants will matter. They already do. The question is whether your store shows up when someone asks.

Matt Harris is the founder of SEOMelon, an AI-powered SEO and AEO optimization tool for Shopify and WooCommerce merchants. He's building in the TinyFish Accelerator.