Get products recommended in ChatGPT & Perplexity shopping

Ask ChatGPT for "the best noise-cancelling headphones under $200," or Perplexity for "a good CRM for a ten-person sales team," and you don't get a page of links to wade through anymore. You get a short list of actual products — often with a price, a line on who each one suits, and a link to buy. That's AI shopping, and it's quietly becoming the new shelf. If your product isn't one of the names the assistant hands back, you're invisible at the exact moment someone has their wallet out.

The reassuring part: being recommended isn't luck. These answers are assembled from data the engines can find and trust — and most of that data is yours to shape. Here's how it works, and what to actually do about it.

What "AI shopping" actually is now

The plumbing differs by engine, but the outcome is the same. ChatGPT surfaces product picks directly inside answers; Perplexity returns products with the sources it pulled from; Gemini ties into Google's shopping graph. In every case the assistant chooses a handful of products for a buying-intent question and, increasingly, shows a price, a rating and a link. The important difference from a Google Shopping ad is that you're not bidding for placement — the engine decides on the strength of what it can read about your product and what everyone else says about it. That makes this earned, not bought, which is good news if you do the work and bad news if you ignore it.

Why a buying query is a harder test than a normal one

For "what is generative engine optimization," a model just needs a clean definition. For "best project management tool for agencies under $20 a seat," it has to put its neck out — name specific products, and ideally attach a price and a reason. That higher bar is exactly why thin product pages get skipped. If your page doesn't state, plainly and in a form a machine can read, what the thing is, what it costs, who it's for and why people rate it, the model can't safely recommend it and moves to a competitor who spelled all of that out. The pages that win read less like a brochure and more like a confident, factual answer to "should I buy this, and why."

Where the engines get their product data

Before you fix anything, it helps to know what these systems are reading:

• Your own product pages — the primary source, especially the structured data in them.
• Product schema — machine-readable price, availability, rating, brand and identifiers.
• Marketplace and retailer listings — Amazon, app stores, directories where your product also lives.
• Reviews and ratings — on your site and on third-party review platforms.
• Comparison and "best of" content — the roundups and versus pages models lean on for shortlists.

How to get your products recommended

1. Add Product structured data

This is the price of entry. Product schema (with Offer and AggregateRating) tells the engine your price, whether you're in stock, your average rating and how many reviews back it, in a format it doesn't have to guess at. A page that hands over "$149, in stock, 4.6 from 812 reviews" as clean data is far easier to place in a shopping answer than one where a model has to scrape a number out of a banner image. For the how-to on schema, see optimizing your website for AI.

2. Write product pages a model can actually quote

Lead with what it is and who it's for, then back it with specifics. State the real use-cases, the concrete specs, and an honest line on where it fits versus where it doesn't. "Best for small agencies that bill hourly" is quotable; "the perfect solution for everyone" is not. The same rules that make blog content citable apply doubly to products — the deeper dive is in how to optimize content for AI search results.

3. Earn reviews and third-party mentions

Models resolve ties by trust, and trust comes from corroboration. A product that's rated on your site, reviewed on a third-party platform, and mentioned in a couple of independent roundups looks safer to recommend than one that only talks about itself. You don't need hundreds of reviews — you need enough credible, consistent signal that the engine believes the product exists and performs as claimed.

4. Create the buying-intent content

The queries that trigger AI shopping are predictable: "best X for Y," "alternatives to Z," "is A better than B for C." If you publish clear, honest content that answers those exact questions — comparison tables, use-case guides, straight versus pages — you give the engine something to pull when the buying question comes up. This is where a lot of the citation actually happens; the assistant lifts from the comparison page, not the homepage.

5. Keep price and stock current

A shopping answer with a wrong price is worse than no answer, so engines favour products whose data looks maintained. If your structured price is stale or your stock status is wrong, you're a risky pick and a fresher competitor gets the slot. Treat price and availability as living data, not a one-time setup.

See which buying queries you actually show up for

You can't improve what you can't see. Before rewriting anything, find out which buying-intent prompts already name your product, which name your competitors instead, and where you're missing entirely. GEOpta's shopping analysis runs those buyer queries across the engines and shows exactly where you stand, so you fix the gaps that matter instead of guessing. Start with a free AI Visibility Score to baseline your brand, then work the list.

Bottom line

AI shopping rewards the same thing good salespeople do: a clear, specific, trustworthy answer to "why this one." Give the engines clean product data, pages that read like confident factual answers, real reviews, the comparison content buyers actually ask for, and prices you keep current — and you become one of the names the assistant returns. Ignore it, and your competitors get recommended at the checkout while you're still optimizing for links. See where you stand with a free AI Visibility Score, then close the gaps.

Frequently asked questions

How do I get my product recommended by ChatGPT and Perplexity?

Give the engines clean, trustworthy data about your product. Add Product and Offer structured data so price, availability and ratings are machine-readable, write product pages that state clearly what the product is and who it's for, earn reviews and third-party mentions for corroboration, publish the comparison and 'best X for Y' content buyers ask for, and keep your price and stock current. Then measure which buying queries name you and fix the gaps.

Is AI shopping the same as Google Shopping ads?

No. Google Shopping ads are paid placements you bid for, while AI shopping recommendations in ChatGPT and Perplexity are earned — the engine chooses products based on what it can read about them and what other credible sources say. You influence AI shopping through structured data, clear product content and reviews rather than an ad budget.

What structured data do I need for AI shopping visibility?

At minimum, Product schema with an Offer (price, currency and availability) and AggregateRating (average score and review count), plus brand and product identifiers where you have them. This lets ChatGPT, Perplexity and Gemini read your price, stock and rating as reliable data instead of guessing from the page, which makes your product much easier to include in a shopping answer.

Why does AI recommend my competitors instead of me?

Usually because their product data is clearer and more corroborated than yours. If a competitor has clean Product schema, specific pages, visible reviews and mentions in comparison content, an engine can recommend them with confidence, while a vague or thinly-reviewed page is a riskier pick. Closing that gap — structured data, specific content and third-party proof — is how you win the spot back.

How do I see which shopping queries my product appears in?

Run those buying-intent prompts across the AI engines and track which name your product, which name competitors, and which miss you entirely. GEOpta's shopping analysis automates this so you can see your product visibility for real buyer queries and prioritise the gaps, rather than checking prompts by hand.