How to Get Your Products Recommended by ChatGPT Shopping

ChatGPT now answers “what should I buy” directly, with a shortlist of products — and it builds that shortlist mostly from structured data, not from whoever ranks #1 on Google. Around 83% of the products ChatGPT shows in shopping carousels come straight from Google Shopping feed data. So getting recommended is less about clever copy and more about three unglamorous things: a complete product feed, crawler access for OAI-SearchBot, and server-rendered product pages with full schema. This guide walks the exact checklist, and flags where some platforms get an unfair head start.

If you’re not sure whether AI engines can even see your store today, run the free AI visibility check first — it tells you in ten seconds whether OAI-SearchBot and the other AI crawlers are allowed in, and whether your pages carry the data they need.

How ChatGPT actually picks products

There are two different things happening, and it’s worth separating them:

1. Organic shopping recommendations — when a user asks “best waterproof work boots under £120”, ChatGPT assembles a shortlist. This is drawn largely from Google Shopping feed data (roughly 83% of carousel products), plus product pages it can crawl, independent buying guides, and reviews.
2. Instant Checkout — buying inside the chat via the Agentic Commerce Protocol. Shopify and Etsy catalogues are already wired in; other platforms have to do that integration themselves.

For most merchants the prize right now is #1: being in the shortlist. You do not need to apply or pay to be eligible for organic recommendations — but you do need your data to be present, accurate and readable. Miss any of the three foundations below and you’re invisible regardless of how good the product is.

1. Fix the feed — it’s doing most of the work

Because the majority of recommended products trace back to Google Shopping data, your feed is the single highest-leverage surface:

• Coverage: every product you want recommended must be in the feed and approved in Google Merchant Center. A product that isn’t in the feed effectively doesn’t exist to the shopping layer.
• Attribute completeness: aim for ~95%+ completion on the attributes that matter — title, description, brand, gtin/mpn, price, availability, color, size, material, product category. Sparse feeds get filtered out of comparisons.
• Accurate price and stock: engines will not surface a product they can’t confirm is available at a stated price. Stale feeds quietly drop you.
• Descriptive titles: Regatta Men's Waterproof Safety Boots, Steel Toe, UK 7–12 beats Work Boots — the feed title is a ranking and matching signal, not a label.

If you only do one thing this month, get the feed to full completion.

2. Let the AI crawlers in

The feed gets you eligible; crawlable pages get you trusted and quoted. AI engines cross-check feed data against your live product pages. If they can’t fetch the page, that confirmation never happens.

• robots.txt must allow OAI-SearchBot (ChatGPT’s shopping/search crawler) — and ideally PerplexityBot, ClaudeBot, Google-Extended and Bingbot too. A single Disallow here is a hard ceiling on visibility. See the AI crawler directory for the exact user-agent strings.
• No bot-challenge pages. If Cloudflare or a WAF rule challenges the crawler’s user agent, the engine sees an empty page — as good as a block. This is the most common silent failure.
• Server-rendered content. Most AI crawlers don’t execute JavaScript. If your product name, price and description only appear after client-side rendering, the crawler sees an empty shell. This is where a lot of modern storefronts quietly lose — see robots.txt for AI crawlers and the note on headless below.

In practice the trap that catches the most stores is subtle: the product page renders server-side, but the price is injected by JavaScript after load — common with configurable products, currency switchers and “from £X” widgets. The owner swears the price is “right there” because they see it in the browser; the crawler fetches the raw HTML and finds no price at all, so the product never gets recommended. Always check the raw source (right-click → View Source, not Inspect), and search it for the actual price string.

3. Give engines structured, quotable product data

Feeds and crawlers get you in the room; schema is what lets an engine state a fact about your product confidently enough to recommend it. Every product page should carry complete JSON-LD Product markup:

• name, description, image, sku, brand
• offers with price, priceCurrency, availability — without a confirmable price, you don’t get recommended
• aggregateRating and review where you have genuine reviews
• gtin/mpn so the engine can match your product to the same item elsewhere

Test it the way an engine would: paste your rendered product HTML into an LLM and ask “what is this product, what does it cost, and is it in stock?” If the model can’t answer cleanly, neither can the shopping engine. (For the Magento-specific version of all this, see the Magento AEO checklist.)

4. Build the buying guides engines learn from

Product data makes you eligible; category buying guides are what engines reason from when they decide which of the eligible products to name. A clear, honest guide — “how to choose X”, one H2 per real buyer question, concrete price bands, genuine “don’t buy this if…” — acts as a training document the engine leans on. Two practical rules:

• Answer the buying question in the first paragraph of category and guide pages, in a quotable sentence. Engines lift these directly.
• Cover the awkward questions buyers ask a chatbot before purchase: sizing, compatibility, returns, warranty, lead times. Mark them up with FAQPage schema.

5. Reviews are a heavier signal than you think

Reviews are one of the strongest inputs an engine uses to choose between comparable products — volume and quality. A product with 300 genuine reviews at 4.6★ gets recommended over an identical one with none. Get reviews syndicated into your schema and, where relevant, onto third-party sites the engines already trust.

The platform head-start problem

Not every store starts from the same line, and it’s worth being honest about it:

• Shopify and Etsy stores are already integrated for the Agentic Commerce Protocol and Instant Checkout — a genuine head start on in-chat buying.
• Magento, WooCommerce and custom stacks appear through their feed and crawlable pages like everyone else, but get no automatic checkout integration and, crucially, often ship product content that only renders in JavaScript — so the crawler confirmation in §2 silently fails.

This is the deciding factor for a lot of stores: if your storefront renders client-side, fix that before anything else, because the feed and schema work is wasted if the engine only ever sees an empty page. A correctly built headless storefront solves it — you get static-fast pages and fully server-rendered HTML that AI crawlers read perfectly, which is the strongest possible foundation for AI shopping visibility. Done wrong (client-side rendering behind a SPA) it makes things dramatically worse. If you’re weighing a rebuild for exactly this reason, Agentic E-Com builds AI-ready headless Magento storefronts with this as the design goal — but verify any before/after with the checker so you’re deciding on evidence, not a pitch.

Your first-week checklist

1. Run the free AI visibility check — confirm OAI-SearchBot and the other five crawlers are allowed.
2. Get your Google Shopping feed to ~95%+ attribute completion, prices and stock accurate.
3. Confirm product pages are server-rendered (view source; is the price in the raw HTML?).
4. Add complete Product JSON-LD with offers, price, availability, gtin.
5. Publish one honest category buying guide with an answer-first opening paragraph.
6. Ask ChatGPT your five “best X to buy” questions today, log which stores it names, and repeat monthly.

This guide is one foundation of the wider AEO for ecommerce playbook — the pillar that covers all six. If the engines still skip you after fixing the feed, work through why ChatGPT doesn’t recommend your store. The overlap with classic SEO and with AEO more broadly is deliberate — everything here also helps your Google rankings and your visibility in Perplexity and Gemini. Start with the feed and crawler access; they’re where the invisible stores lose.