Guides

ChatGPT Ads for Ecommerce Stores

Product feed ads are the one format OpenAI says performs best. Here is what your store can run, what the results show so far and the setup to start with.

By the Launch GPT Ads team · October 20, 2026

The short version

Your store can advertise most consumer goods on ChatGPT. The strongest format is the product feed, which turns your catalog into ads with image, name, price and rating, and OpenAI says feed ads have been among its best performers. The published wins come from OpenAI and its partners, such as Newegg at 3 times return on ad spend. Many small stores that posted their own tests got clicks and no sales. Start with products people research before they buy, a clean feed and working conversion tracking.

What your store can advertise

  • Allowed in all 63 self-serve countries: household and consumer goods, including fitness equipment and wearables sold without medical claims.
  • Allowed in a normal retail setting: lingerie, underwear and swimwear. Sexual health products are banned.
  • Banned: alcohol above 0.5%, tobacco, vaping and nicotine, counterfeit goods, and products with THC. Hemp products such as CBD creams with no THC may be allowed.
  • Health products, US only: dietary supplements and consumer medical devices are allowed in the US case by case. Supplements with unsupported health claims sold outside major retail chains are banned everywhere.
  • Banned in any category: products that play on body-image worries.

Your ad and your landing page also have to match. OpenAI's own example of a breach is a food delivery ad that leads to an alcohol delivery page.

What the results look like so far

OpenAI has published these store results, each measured by the store or by a measurement company:

  • Newegg, an electronics retailer: 3 times return on ad spend across its campaigns over 28 days, and 7 times during a 2 week sale.
  • Portland Leather Goods, leather bags: 93% of the visitors from ChatGPT ads were new to the store, measured by Triple Whale.
  • Dose, a wellness brand: 67% of the extra purchases came from new customers. The study found 2.3 times more extra orders than last-click tracking had counted.
  • Fospha's advertisers: 83% more revenue per 1,000 ad views than their paid social average.

The measurement companies sell their tools, and OpenAI chose which results to publish, so read these as the best cases.

Stores that posted their own numbers show a wider spread. Opascope, an agency, reported about $60,000 of spend over 15 days for 1.49 times return on ad spend, at $1.72 a click and a 2.35% conversion rate, with single days ranging from 0.2 to 2.9. One store owner posted $1,000 of spend, 517 clicks and 1 sale. A Spanish supplement brand spent €37.61 for 160 clicks and no sales. Members of the eComFuel forum for store owners describe good click rates and almost no conversions so far. OpenAI's Shopify app has a rating of 2.2 out of 5 from 21 reviews, and most complaints are about products that did not sync.

Big retailers are still testing too. Best Buy was the largest ChatGPT advertiser in February 2026 and showed no ads at all by mid-May, according to Sensor Tower. From March to August, retail and consumer brands had 10.4% of US ChatGPT ad views, from 1,510 advertisers led by Home Depot.

Where your products can show up

In OpenAI's study of ChatGPT messages from mid-2024 to mid-2025, 2.1% asked about products to buy, such as "Recommend a good laptop under $1000". Another 5.7% were about health, fitness, beauty or self-care.

When OtterlyAI ran a fixed set of US shopping questions in July 2026, 79.2% of the retail answers carried an ad. And Adobe found that visitors who reached US retail sites from ChatGPT's normal answers in July 2026 converted 60% better than other traffic. Those are free answers, not ads, but they show that people already shop through ChatGPT.

The setup to start with

  1. Let OpenAI's crawlers in. OAI-AdsBot has to reach your product pages and OAI-SearchBot has to reach your product images. If your image host blocks it, the whole feed fails.
  2. Connect your product feed. Upload a file, or better, a hosted link that updates itself, because feed items expire after 2 weeks. On Shopify in the US, OpenAI's app brings in your catalog and sets up tracking. Check that every product shows up, because syncing is the most common complaint.
  3. Track sales. Use the OpenAI pixel together with the Conversions API, and send the order value with every purchase. ChatGPT's return on ad spend column only counts the purchase event.
  4. Choose the Conversions objective, billed on views. This is OpenAI's own recommendation, because it learns from people who buy after seeing the ad as well as those who click. Of the people who bought within a day of seeing an ad without clicking it, 52.7% bought within the first hour.
  5. Split your catalog into ad groups by theme, and describe each one as the situation a buyer is in. OpenAI's example is "Cushioned everyday running shoes for beginners training for their first 5K".
  6. Keep one feed campaign running all the time and add a separate campaign for each sale. That is the structure Newegg used.
  7. Decide on creative enhancements before October 29, 2026. From that date OpenAI can rewrite your titles, change your landing page and translate your ads unless you switch it off.

Feed campaigns can only target whole countries for now. Standard campaigns can target states and cities.

What to watch

  • The click price against your margin. With Opascope's numbers, $1.72 a click and 2.35% of clicks buying, each sale costs about $73 in clicks. A $20 product cannot carry that. A $300 product can.
  • Clicks that do not show up in your analytics. One agency saw about 100 clicks in ChatGPT and 20 in its own tracking. Count sales in your store, not only in Ads Manager.
  • Slow reports. Spend shows about 7 to 8 hours late and conversions 24 to 48 hours late, so wait before you change anything.
  • Europe. Ads in Europe use only the conversation and the person's general location, with no personal data, so your product titles and context hints do all the matching.

Which stores should test first

  • Good fit: products people compare and ask about before buying, such as electronics, furniture, gear and higher-priced items, with an order value that can carry a few dollars a click and a clean product feed.
  • Weaker fit: cheap impulse products, products sold on health claims, and anything in a banned category.

The smallest test ChatGPT allows is $25 a day, up to $175 a week. Opascope recommends $5,000 to $10,000 for a test you can trust. Whatever you spend, judge it on what each sale cost you, next to your Google Shopping and Meta numbers for the same weeks.

Based on each platform's own help pages and on the studies named below, as we read them on October 20, 2026. Prices and rules change often, so check the current numbers before you spend.

Sources: OpenAI ad policies v1.6, September 10, 2026. OpenAI Help Center, ChatGPT Ads collection (product feeds, conversion measurement, crawlers, creative enhancements, availability). OpenAI customer story (Newegg). OpenAI Ads blog, More ways to measure ChatGPT Ads, October 5, 2026. OpenAI and NBER, How People Use ChatGPT, September 2025. Opascope, ChatGPT ads benchmarks, June 2026. Public posts by store owners on X and Reddit, July to September 2026. Shopify App Store listing for ChatGPT Ads, October 2026. Sensor Tower data in The AI Ad Economy, June 11, 2026. Graphite and Similarweb, The State of ChatGPT Ads, via PPC Land, October 4, 2026. OtterlyAI ChatGPT ads study, July 2026. Adobe data via Digital Commerce 360, August 19, 2026. Digiday, October 1, 2026.

Find Out if ChatGPT Ads Fit Your Business

Tell us what you sell on a short call. We show you how ChatGPT ads would work for it and what a first test would need.

Talk to us
© 2026 Mez Digital LTDHomeGuidesPrivacy PolicyTerms