← All posts
June 23, 2026

Why AI agents recommend some products and skip yours

Ask ChatGPT, Gemini, or Claude for "the best [something] under $X" and you'll get specific recommendations — named products, sometimes named stores. Now ask one about your category. If your products aren't in the answer, it's worth understanding why, because the logic is knowable and mostly fixable.

How the decision actually gets made

AI assistants don't recommend based on your ad budget or a clever SEO trick. They recommend what they can confidently understand. A few things drive it:

Structured, machine-readable data. Agents read product feeds and structured markup — not your hero image or your tagline. If a product's key facts (specs, price, availability, attributes) are cleanly structured, an agent can use them. Buried in prose or baked into an image, they often can't.

Completeness and specificity. To recommend a product for a person, the agent has to know who it's for. "Lightweight all-mountain ski, intermediate, 165–175cm" is recommendable. "Great ski, ride in style" is not. A gap in your data is a gap in your visibility.

Trust signals. Agents check that you're a real, safe place to buy — clear shipping and return policies, reviews, pages that work. They're cautious about sending someone to a store they can't vouch for.

Answering the real question. Buyers ask "which one is right for me?" Pages that actually answer that — comparisons, fit guidance, use-case framing — get surfaced. Keyword-stuffed pages don't.

And the simplest rule of all: you can't be recommended for something the AI can't tell you sell. Absence of data is absence from the answer.

First, can the AI even see you?

Before any of that: many stores accidentally block AI from reading them at all. Two quick checks. Is your content actually in the page's HTML, or only loaded by JavaScript that a crawler may never run? And does your robots file welcome the AI crawlers — the ones from OpenAI, Google, Anthropic and others — rather than block them? If an agent can't read your pages, nothing else on this list matters.

What to do about it

  1. Make product data complete and specific. Add fit, sizing, who-it's-for, and what-it-pairs-with — to your best sellers first.
  2. Add structured data. Product schema (JSON-LD) lets machines parse your catalog reliably. Most platforms support it; Shopify largely handles it when your fields are actually filled in.
  3. Make sure you're crawlable. Server-rendered content, a sitemap, and a robots file that allows AI crawlers.
  4. Fill policies and FAQs. So an agent can confidently vouch for you.
  5. Write for the question. Replace "premium quality" with the specifics a buyer actually weighs.

A quick test you can run today

Want to know where you stand? Ask the three big assistants yourself: "What do you know about [your store]?" and "Recommend a [your category] for [a specific kind of buyer]." For a lot of small shops the honest answer is that you simply don't come up — which is exactly the gap worth closing. (If you'd rather not dig through three chat windows, we'll send you a read on how each one currently sees your store.)

The deepest version of this — an agent that carries your full category expertise and walks each shopper to the right pick — is what Sqrly is built to give every store. But the steps above are free, and they're the foundation either way.

The bottom line

Being recommended by AI isn't luck or budget. It's legibility: making what you sell, and who it's for, something a machine can understand and trust. The stores that do that are the ones showing up in the answer.

Sqrly helps online retailers become the store AI recommends — by giving every shopper and every agent a knowledgeable, on-brand conversation. Join the waitlist.