SEO, AEO and shopping agents: three ways shoppers find products now

Search engine optimization is about ranking in Google's list of links. The shopper does the work: they search, open a few pages, compare and decide. You win with pages that answer the search, a fast and well-structured site, and other sites linking to you.

SEO still matters. But it's built for a shopper who browses, and fewer shoppers browse now.

Answer engine optimization (some call it GEO, generative engine optimization) is about what AI assistants say when a shopper asks them. ChatGPT, Claude, Gemini, Grok and Perplexity don't hand back ten links. They answer: "Get the Front Range. Here's why." Two or three products get named. Everyone else doesn't exist.

Assistants decide by reading the web as they answer: review roundups, comparison articles, forums, and brands' own pages. So AEO comes down to two things. Be in the pages they read, and make your own pages say plainly what you sell, who it's for, and why it's better.

Answers also vary from one ask to the next, so the honest measure isn't a rank. It's how often each assistant recommends you, and which sites it leaned on to decide.

Shopping agents go a step further: they find products and buy them. Rather than reading articles, they search a product catalog. Shopify's shop.dev is the catalog agents use to search Shopify stores, and it ranks listings on the product data itself.

That makes this the most controllable of the three. There's no editor to win over. Your position depends on your titles, descriptions, details, ratings, price and stock, all of which live in your Shopify admin.

Know your space first: the searches your customers run to find what you sell without naming you. Every channel above is a contest over those same searches.

Then fix what you control, in order: your product listings (shopping agents read them directly), your own pages (assistants and Google both read them), then the outside pages assistants cite. Bark Twice measures the second and third channels for every brand it tracks, and turns each finding into a specific fix with what it's worth.