ChatGPT Shopping: What It Is & How Products Appear

We’re moving insanely fast that if someone had told me about ChatGPT Shopping just a month ago, I would’ve thought it was early.

But just now, I placed my third order within a week through ChatGPT Shopping – to compare just how useful this is, I can’t even remember the last time I used Google Shopping.

The initial results clearly show that many people, like me, were already ready to use ChatGPT for shopping.

I hope that by the time you finish this article, you’ll see why I said this — and how this new feature opens up exciting possibilities for brands.

But first, we need to discuss in detail what this feature is all about.

Key Takeaways:

• Product discovery has been moving toward intent and context for a while; ChatGPT Shopping is accelerating this trend.

• It transforms user shopping behavior towards conversation and dialogue-based exploration, enabling product discovery and purchase within a chat experience and thus creating a richer shopping journey.

• It provides purpose-driven product recommendations based on remembered preferences and product reviews from across the web – not just general popularity.

• Instead of a long list of options, it offers a carefully curated shortlist of high-quality suggestions.

• By adding visuals, reviews, and direct links into the conversation, it smooths the entire path from discovery to purchase.

• It is the AI itself that determines which products are displayed. These developments are forcing brands and retailers to rethink how to stay visible and capture attention in this new shopping environment – and face urgent tasks like implementing answer engine optimization (AEO).

What is ChatGPT Shopping?

The ChatGPT Shopping feature lets you easily discover products from various sellers across the internet.

It presents shopping cards that include details such as product ratings, prices, sellers, and top picks.

As of April 28, 2025, it’s available across all versions of ChatGPT.

ChatGPT Shopping Product Cards

Each product card is essentially a “shopping recommendation” and includes:

• Images

• Simplified titles and descriptions

• Price (from the listed seller)

• Links to product pages

• AI-generated tags: “Best for Espresso,” “Rich & Bold Flavor,” “Budget-friendly,” “Most Popular,” “Top Pick,” “Best Value,” or “Most Comfortable”

• Star ratings and number of reviews from third-party sources

• AI-generated review summaries

What’s New:

Maybe, like me, you were already using ChatGPT to research and buy products.

But when you searched for a product in ChatGPT before, the recommendations appeared only as bullet-pointed lists. There was little context or reasoning shown behind each item.

Thanks to the Shopping feature, ChatGPT now offers a much more advanced and enriched shopping experience to users with buying intent.

Now, several elements that simplify decision-making are presented in an easy-to-read format:

Interactive product cards with visuals, real-time pricing, star ratings, review context, and direct links to third-party retail sites for completing purchases…

The Core Shift: Our Shopping Behavior

The product cards integrated into ChatGPT are an important innovation.

But even more important is this: ChatGPT Shopping provides you with ad-free, organic (for now) product recommendations tailored to your preferences based on previous chats and queries.

In fact, the most significant shift with ChatGPT Shopping is our shopping behavior itself. We are now shopping by chatting with an AI that knows us. ChatGPT goes beyond just answering questions – it shapes how shoppers discover products.

This personalized experience offered by ChatGPT sets it apart from other answer engines. For example, while Perplexity’s Buy with Pro initiative is command-based, ChatGPT provides memory- and context-based recommendations. This makes the suggested products far more personalized and relevant.

Though it’s still early days, it already plays a role in changing how millions of people make buying decisions. This core shift introduces new challenges and opportunities for brands in terms of how their products are represented and how they measure that impact.

✨ Vision: Currently, you still need to go to the seller’s website to complete the purchase. But inevitably, in the near future, once direct payment within ChatGPT is enabled, you’ll be able to complete a purchase with just a verbal instruction.

Why Does ChatGPT Shopping Matter?

And now, let’s get to the most critical part – especially for marketers and business owners.

ChatGPT now visibly links to third-party shopping sites.

In clearer terms: ChatGPT can drive traffic to your e-commerce site.

Essentially, this feature is similar to Google Shopping. The logic is pretty simple: It detects if a query is shopping-intent-based and shows product options in the answer screen.

But there’s one major difference:

This time, the query happens inside an answer engine.

That means ChatGPT delivers precisely targeted product suggestions.

To understand how this could reshape how consumers discover and purchase products online – and thus transform the e-commerce industry – just imagine ChatGPT processing a fine-tuned query like this within the answer box:

“I’m looking for a smart bracelet for health-conscious, active adults that can track daily activity and monitor sleep quality. It should have a silicone or TPU strap, be adjustable to different wrist sizes, and preferably be black in color. It needs to be IP67 dust and water resistant and lightweight for all-day comfort. Additional features like heart rate monitoring, step counting, sleep tracking, and phone notifications are a plus. I prefer brands like Xiaomi or Fitbit, but accurate data tracking and long battery life are my top priorities. My budget is between 200 USD and 400 USD. I want to buy it from reliable online tech stores. It should be compatible with both Android and iOS. Please find the most suitable options and provide a short description, key features list, and if possible, highlight key user review points for each product.”

In a simpler framework… If a user’s intent is to find a product under a certain price or from a provider with trustworthy return policies, ChatGPT will surface products that meet these criteria. The user can then follow up with more questions to see extra images or find out what others are saying about the item.

🧪 Formula:

A customer who knows what they want + An answer engine that knows what to deliver = Conversion 🏅

👩‍💻 Pro tip: If you’re thinking, “But not everyone is a prompt expert – they might not express themselves this clearly,” keep reading. We’ll explain how this works even with poor prompts.

Advantages of ChatGPT Shopping for Brands

This new feature from ChatGPT makes it easier than ever to reach your market.

Here’s why:

• It answers over 1 billion queries per week, and competition for answer engine optimization is still low.

• People are actively using it to research a wide range of shopping categories like beauty, home goods, and electronics.

• You don’t need to pay to be discovered.

• Optimizing for ChatGPT Shopping — before it’s too late — gives you a first-mover and competitive advantage.

• When you optimize for ChatGPT, your visibility often increases in other answer engines like Perplexity and CoPilot too (since they follow similar patterns).

• Your previous strategic SEO efforts may already align with AEO, and your website may already be ChatGPT-ready.

How ChatGPT Chooses Recommendations: What We Know

The AEO Radar team has spent the last 10 days testing hundreds of shopping queries and reached a clear conclusion:

ChatGPT’s shopping results differ significantly from those of Google Shopping and Bing Shopping.

In fact, it sometimes does things that can be considered radical—like omitting some global brands entirely and prioritizing local businesses.

According to OpenAI, ChatGPT shows products that it predicts will meet the user’s intent and need.

But how does it understand the user’s (shopping) intent and select the products?

Here are the conclusions our team reached after analyzing OpenAI’s statements and the early practical implementations:

1. It Uses What It Knows About You

ChatGPT’s memory, available to Plus and Pro users, plays a key role in product selection.

This feature allows ChatGPT to recall useful details across all your conversations.

So even if a user types in an incomplete or vague query, the answer engine may still capture the correct context for their shopping intent.

For example, a user who owns a Golden Retriever may have discussed various topics with ChatGPT over time.

Even if the user forgets those conversations, ChatGPT might still remember the following when making a product recommendation:

- The user owns a 6-month-old Golden Retriever puppy named Goldie.

- Requests for help with basic training may indicate the owner is new to dog parenting or has limited experience with this breed.

- Mentioning hip dysplasia risk suggests the dog may be genetically predisposed to it or the owner is informed about it.

- Issues with repetitive barking and reactivity around other dogs indicate behavioral tendencies.

- Travel plans and interest in dog sports imply an active lifestyle with the dog.

As a result, when this user searches for leashes, ChatGPT might recommend anti-bark collars or reward-based training toys as well.

However, because the model interprets user intent, it may sometimes make mistakes—resulting in suggestions you don’t like.

Even a one-time, unrelated command you gave earlier might influence how ChatGPT recommends a product.

Still, you can easily clarify your preferences with another command, and ChatGPT will ignore them moving forward.

For instance, if you say you’re not interested in training toys, the model won’t suggest them again in future queries.

🗒️ Quick note: This feature may not be available in some countries or disabled by default for legal reasons, which means your shopping experience may not be as rich and powerful as described here.

2. It Uses Your Custom Instructions

If you previously added custom instructions to ChatGPT, the answer engine factors them into its product suggestions.

This feature is increasingly used, and as ChatGPT Shopping becomes more common, it’s likely people will include preferences like “I want affordable options” in their instructions.

Therefore, it’s inevitable that ChatGPT will give more weight to these instructions than anything else when filtering product suggestions.

👩‍💻 Pro tip: Understanding what kinds of shopping instructions your potential customers have is extremely valuable for answer engine optimization efforts.

3. It Puts User Needs at the Center

Never forget. This system has one goal:

To accurately understand the user’s need and recommend the most suitable products.

That’s why it suggests products not just by technical specs, but based on usage intent and real-life scenarios.

How Does It Do This

Consider this example:

User query:

“I’m looking for bluetooth headphones with good noise isolation that I can use while working out at home.”

1. Intent Identification: Professional-Level Analysis

a. Stated Primary Goal:

“Bluetooth headphones for home workouts.”

b. Needs (Functional and Emotional):

Functional:

• Wireless connectivity (stable Bluetooth)

• Secure fit (ergonomic design)

• Noise isolation (active or passive)

• Sweat and water resistance

• Sufficient battery life

• Microphone quality (for potential calls)

Emotional:

• Comfortable use → helps with focus

• High sound quality → keeps motivation up

• Stylish design → feel-good factor

c. Implicit or Hidden Needs:

• Sudden movements during exercise → headphones must stay secure

• Others may be in the house → strong noise isolation is crucial

• Likely to listen to intense music → bass balance matters

• No wires = need for freedom of movement

• Noisy environment = need for ANC (Active Noise Cancellation)

2. Filtering: Deep Inquiry-Based Approach

High-level filtering questions the system asks itself:

Comfort:

• Do the headphones stay in place during treadmill runs or HIIT workouts?

• Do users report discomfort after long use?

• How lightweight is the product?

Durability & Suitability:

• What’s the IP rating (e.g., IPX4, IPX5)?

• Tested for sweat and dust resistance?

• Can it handle drops and intense movement?

Sound Performance:

• Is ANC effective in real-time workout environments?

• Does it support sound profiles suitable for preferred music genres (e.g., electronic, hip-hop)?

• How well does it isolate external noise?

Connectivity:

• Does Bluetooth disconnect during workouts?

• What Bluetooth version does it use (5.0, 5.3)?

Battery Life:

• How long does it last for a 1-hour session?

• What’s the total battery life with the case?

User Reviews & Scenario Fit:

• Do reviews include phrases like “ideal for workouts,” “I use it at home,” or “stays in place”?

• What do most complaints focus on (sound, comfort, connectivity)?

3. Scoring Criteria for Perfect Recommendation

The system creates a context-specific, weighted score based on all the above data.

4. Result: A Real Scenario-Based Product Recommendation

Jabra Elite 7 Active is built for tough workouts with its IP57 rating and ShakeGrip coating that prevents slipping. The ANC feature significantly reduces background noise. With a 9-hour battery life, it’s ideal for long sessions. User reviews frequently mention peak performance during at-home workouts and running.

Why Understanding the Background Matters

In the last example, the headphone isn’t recommended because of “great audio,” but because it fits the real-life scenario of the user.

The product is positioned more like a “workout partner”—a solution aligned with life, not just tech specs.

But what if there’s a piece of data from the user’s past behavior that they’ve forgotten—but it could drastically affect the outcome?

For instance:

The user previously mentioned ear pain with in-ear headphones. They also frequently join video calls during the day and want a headset for more than just workouts.

ChatGPT doesn’t forget that. This directly influences the recommendation and changes the product.

Impact of This Info on the Decision:

1. Ergonomics and Material Choice:

Tight-fitting silicone tips may cause pain → on-ear or semi-open in-ear models should be preferred.

Coatings like “ShakeGrip” may worsen discomfort → softer materials and lighter designs should be prioritized.

2. Multi-Purpose Use:

If it’s also for video calls → microphone quality becomes critical.

Dual-mic tech, rarely found in workout headphones, becomes essential.

Revised Recommendation:

Sony LinkBuds S fits snugly but remains ultra-light, preventing ear pain during extended use. ANC effectively blocks out noise during workouts, while the ambient mode allows users to hear others at home. Its crystal-clear microphone quality makes it ideal not just for workouts but also for Zoom calls. Considering the user’s past experience with ear pain, this model offers a well-balanced solution in terms of comfort and performance.

How Are Merchants Chosen?

If you look at what the ChatGPT side says, you’ll understand that the merchant data in product cards belong to third-party providers.

In other words, ChatGPT does not use a ranking algorithm.

But don’t get this wrong: it doesn’t rank, but it does filter.

So, while there is no ranking algorithm, there is a filtering and recommendation algorithm — and this algorithm is decisive.

When other sellers are filtered out, ChatGPT might highlight you and recommend your products to users.

This does not mean listing all options or sorting by price or return policy — it means selecting and showing a few.

In fact, if ChatGPT had left such an important decision to third parties, it wouldn’t be able to offer a tailored experience.

Let me share a secret with you at this point:

Based on our experience, ChatGPT doesn’t choose sellers — it chooses products.

Then, it rewards the sellers who make that product easy to select.

Because unless a product meets the perceived user need, the seller doesn’t matter.

However, in general and vague prompts, it’s possible to encounter a price-based ranking. That’s because the third-party platform from which ChatGPT pulls data might use a price-based default sorting system.

How to Appear in ChatGPT Shopping Results: Best Practices for Brands

Any merchant’s products can appear in a ChatGPT conversation.

Even if your products aren’t listed on Google Shopping or have limited visibility there, you still have a chance.

That’s because product results are notably different from Google Shopping — this is where the special algorithms come into play.

Here are some proven tips to ensure your products appear in ChatGPT shopping results:

1. Allow ChatGPT Crawlers

To gather information from the web, ChatGPT uses the OAI-SearchBot.

If you block this crawler on your site, your products won’t appear in ChatGPT’s product cards and won’t be suggested by ChatGPT.

You can check your robots.txt file to ensure OAI-SearchBot has uninterrupted access.

If your robots.txt file does not block OAI-SearchBot (which is the default), no additional optimization is required.

2. Use Structured Data

ChatGPT requires structured data feeds that include product details in order to display your products.

If your data isn’t structured, ChatGPT can’t see it, discover your products, interpret them, or recommend them — it will behave as if they don’t exist.

Schema markup — especially in JSON-LD format — is the most common, simple, and effective way to structure your data.

Here are examples of unstructured product data content:

id: "SKU67890"
A unique product identifier.

title: "Adidas Climacool Running Tights"
The product name; short, descriptive, and searchable.

link: "https://example.com/product/adidas-climacool-tights"
Direct product page URL.

image_link: "https://example.com/images/tights-front.jpg"
Main product image URL, usually front view.

price: "$599.90"
Current sale price.

description: "Ideal performance tights for long runs with breathable design."
A short, original, persuasive product summary.

availability: "in stock"
Stock status: ‘in stock’, ‘out of stock’, ‘preorder’, etc.

condition: "new"
Product condition: ‘new’, ‘refurbished’, ‘used’, etc.

brand: "Adidas"
Brand name.

gtin: "04054321567891"
Global Trade Item Number (such as EAN/UPC).

mpn: "AD-TIGHT-2025"
Manufacturer part number.

multipack: "1"
If purchase includes multiple items, this number is shown.

is_bundle: "no"
Is it sold as a bundle? Yes/No.

color: "Navy Blue"
Primary color of the product.

size: "M"
Size information.

age_group: "adult"
Target age group: ‘infant’, ‘kids’, ‘adult’.

gender: "female"
Target gender: ‘male’, ‘female’, ‘unisex’.

item_group_id: "TIGHTS2025"
Common ID for grouping product variants (color, size, etc.).

shipping: "Free shipping in France"
Shipping info; e.g., “Free shipping in the US”, “Shipping paid by buyer”.

tax: "VAT included"
Tax info; whether VAT or other sales taxes are included.

pattern: "solid"
Product pattern: ‘striped’, ‘polka dot’, ‘solid’, etc.

material: "recycled polyester"
Primary material info.

additional_image_link:
["https://example.com/images/tights-back.jpg", "https://example.com/images/tights-detail.jpg"]
Additional images showing different angles, details, usage scenarios.

size_type: "standard"
Size type: ‘plus size’, ‘petite’, ‘slim fit’, ‘maternity’, etc.

size_system: "EU"
Size system standard: ‘US’, ‘UK’, ‘EU’, etc.

canonical_link: "https://example.com/product/adidas-climacool-tights"
Canonical URL for engines to avoid duplicate content issues.

Now, let’s convert this data into a JSON-LD format for inclusion in structured data schemas using Schema.org’s Product type (@type: Product):

<script type="application/ld+json">
{
  "@context": "https://schema.org/",
  "@type": "Product",
  "name": "Adidas Climacool Running Tights",
  "image": [
    "https://example.com/images/tights-front.jpg",
    "https://example.com/images/tights-back.jpg",
    "https://example.com/images/tights-detail.jpg"
  ],
  "description": "Ideal performance tights for long runs with breathable design.",
  "sku": "SKU67890",
  "mpn": "AD-TIGHT-2025",
  "brand": {
    "@type": "Brand",
    "name": "Adidas"
  },
  "offers": {
    "@type": "Offer",
    "url": "https://example.com/product/adidas-climacool-tights",
    "priceCurrency": "TRY",
    "price": "599.90",
    "priceValidUntil": "2025-12-31",
    "itemCondition": "https://schema.org/NewCondition",
    "availability": "https://schema.org/InStock"
  }
}
</script>

Your structured data is now ready to be used on your website so that product information can appear in ChatGPT.

However, the two points we’ve outlined so far are not enough on their own. The real difference comes from how well you optimize your product attributes.

ChatGPT crawler access and structured data put your products in the game; optimization puts them at the top of the results.

🗒️ Quick note: The provided example is for use on your website only. Once OpenAI begins accepting feed files directly, you may need to convert your data to XML, TSV, or CSV formats.

3. Highlight the Benefits of Your Products

Answer engines like ChatGPT prefer clear and reliable answers.

In line with this preference, lengthy, generic product pages that don’t address a user’s need map won’t perform well.

Contrary to popular belief, in this new game, technical descriptions won’t give you an edge.

What will earn you clarity and authority is clearly explaining the real-world benefits and usage scenarios of your product.

Because if you don’t explain and emphasize this properly, ChatGPT won’t understand what your product is for.

At this point, two scenarios emerge:

1) Your product already meets a refined need: Emphasize its unique benefits.

• Why: The user already knows what they want. The only task is to quickly show why your product is better than the rest.

• AEO Strategy: Clear title, benefit-focused description, content supported with structured data.

2) Your product meets an optional need: Highlight its benefits in a unique way.

• Why: The user isn’t actively searching for it, and it doesn’t meet a refined need — which means competition is high. To stand out, you must capture attention with a unique narrative, build emotional or practical relevance, and trigger interest through differentiation.

• AEO Strategy: Use creative and curiosity-driven descriptions that connect with questions (e.g., “Why did I choose this product?”), comparative explanations, and presentations that highlight different usage contexts.

The more richly you describe your product with its unique benefits and usage scenarios, the higher the chances it will be discovered and turned into an answer.

To really connect with consumers and guide their choices, try diversifying your content by highlighting the benefits, explaining its purpose, and showing them how to use it.

And for good AEO, make sure to do content optimization.

Words that are either missing or unnecessarily added can reduce your product’s discoverability to zero.

At the same time, even a small tweak to your content can sometimes be the key that unlocks discoverability in answer engines.

The rules of the ChatGPT Shopping game are directly tied to content optimization.

4. Encourage Product Reviews

Our limited experience with this new setup clearly shows one thing:

If ChatGPT Shopping had a fuel, that fuel would be review content.

Our initial results prove that products referenced in reviews deemed valuable by ChatGPT are prominently featured. (We are currently working on a case study that includes the details.)

Adam Fry from OpenAI confirms this:

“It’s trying to understand how people reviewed it, how they talked about it, what the pros and cons were,”

These statements confirm that reviews are a powerful tool to persuade ChatGPT.

Moreover, it doesn’t matter who wrote them: the brand itself, an editorial publisher, or the user.

Answer engines pull information from a mix of online sources, including user-generated content on communities like Reddit.

So, you should invest in review generation strategies—such as automated email drip campaigns—to increase the number (and quality) of reviews for your products.

Or launch campaigns that reward user-generated content best explaining your product’s unique benefit.

🔑 Key takeaway: ChatGPT never ignores high-quality, well-thought-out reviews of products extensively tested by real people.

🗝️ Key finding: Our experiences show that reviews which are detailed and emphasize benefits around a clear answer are far more valued than general user comments. On the other hand, reviews reflecting a general negative consensus (not individual negative reviews) tend to not be recommended by ChatGPT.

👩‍💻 Pro tip: If your potential users might prompt something like “source Reddit reviews” or prioritize certain types of reviews, do not ignore this possibility. Increasing the quality of your content in those spaces may yield better outcomes.

5. Create ‘Personas with Memory’

A persona is an imaginary character that represents different types of people within our target audience. We create these characters based on real user research.

Our goal is to understand their needs, goals, and challenges in a concrete way, instead of an abstract ‘user’ group. This allows us to better shape our products and services according to these different user types.

Now, it’s a bit different. Thanks to its memory feature, ChatGPT doesn’t forget what you’ve talked about.

So, to get the best results, you need to create personas with memory and shape your content based on these personas.

You can generate multiple personas across different chat windows and try to understand how well these personas align with your product.

6. Ensure Cross-Channel Consistency

ChatGPT does not rely on a single channel.

It learns from various online sources like your product feed and third-party reviews.

Therefore, your digital presence must be consistent not just in isolated parts (e.g., a blog post) but across the entire web.

No matter where your product appears, a holistic strategy is needed to ensure real consistency.

You must even manage inconsistent content on marketplaces and communities you don’t control.

7. Track and Evaluate Your Performance

When ChatGPT Shopping links to websites in shopping results, it usually adds UTM parameters like utm_source=chatgpt.com.

These parameters help you track the traffic source.

With free tools like Google Analytics 4 (GA4), you can monitor this traffic at no cost.

On the other hand, some professional AEO tools allow you to track and analyze:

• How your products appear in ChatGPT shopping results

• The shopping-trigger queries that prompt ChatGPT to display your products

• How your products are positioned and highlighted

• The key performance drivers: which review, comment, retailer, or web source led to better results

• Trends in how often your products are displayed

• Insights into where your competitors are winning

• Identification of gaps, such as weak regions or underperforming retailers

8. Pre-Register for ChatGPT Product Feeds

Currently, sellers cannot directly submit product feeds to OpenAI.

However, the company is working on a system similar to Google Merchant Center.

This will allow sellers to provide and update direct data to ChatGPT for shopping results.

To get notified when you can directly submit your product feed to ChatGPT, fill out the interest form.

Final Thoughts: Brands Must Be Here

To be clear, ChatGPT is not changing how we shop. It is responding very well to the already-changing way we shop.

In this sense, it won’t replace Amazon overnight.

But we’re talking about a disruptive behavioral shift. Over time, it wouldn’t be surprising if these statistics reversed.

As AI agents become more capable, users will prefer frictionless, personalized, and context-aware conversational experiences.

That’s why brands must act today—understand where the customer is heading, and meet them along the path, not just at the end.

FAQS