How Do OpenAI Product Feed Campaigns Work? A Guide for E-commerce Brands

| Zafer Kavaklı

How Do OpenAI Product Feed Campaigns Work? A Guide for E-commerce Brands

Product feed campaigns allow brands to upload their product catalogs to OpenAI Ads Manager and create ads using the products in those catalogs.

This structure is especially important for brands with large product catalogs. Instead of creating individual ads for hundreds or thousands of products, brands can build a more scalable advertising structure through a product feed.

Product feed campaigns can be compared to the logic behind Google Shopping or Meta Catalog Ads. However, because user intent within ChatGPT is more conversational, it is not enough for product data to be technically accurate. Titles, descriptions, prices, availability, images, and category information should be clear enough to address the user’s needs.

What is an OpenAI Product Feed campaign?

An OpenAI Product Feed campaign is a campaign type that allows e-commerce brands to upload their product catalogs to Ads Manager and create ads from those catalogs.

In this campaign model, product information is retrieved from the feed. Fields such as the ad headline, description, image, price, availability, and landing page are directly connected to product data.

This structure provides brands with three main advantages:

  • Turning large product catalogs into ads more easily
  • Improving ad quality by keeping product information up to date
  • Promoting relevant products during purchase-oriented conversations within ChatGPT

Product feed campaigns offer a more sustainable structure than manual ad management, especially for e-commerce brands whose price, availability, variant, and promotional information changes frequently.

OpenAI Ads Product Feed Campaign

How do OpenAI Product Feed campaigns work?

A Product Feed campaign in OpenAI Ads Manager generally follows this process:

  1. A product feed is created in Ads Manager.
  2. SFTP connection details are obtained.
  3. The product feed is prepared according to OpenAI’s supported formats.
  4. The feed is uploaded through SFTP using Optifeed or through an integration developed by the technical team.
  5. Ads Manager processes the feed and lists the products.
  6. The Product Feed campaign type is selected from the Campaigns section.
  7. An Ad Group is created, and either a filtered product set or the entire catalog is selected.
  8. An ad template is prepared.
  9. The preview is reviewed.
  10. The campaign is submitted for review and launched.

This process allows product data and the advertising structure to work together. Campaign success therefore depends not only on media budget, but also on feed quality.

OpenAI Product Feed SFTP

Feed creation and SFTP upload process

To use a product feed in OpenAI Ads Manager, a feed must first be created within the platform. Once the feed is created, the required SFTP connection details can be obtained from Ads Manager.

The SFTP connection is used to upload your product feed securely to OpenAI.

The following points should be considered at this stage:

  • The feed file should comply with OpenAI feed specifications.
  • We recommend updating the file at least once per hour.
  • The product feed should include all required fields and important optional fields.
  • Price, availability, and product status should be kept up to date.
  • Product URLs should be accessible.
  • Images should work correctly and be high quality.
  • Products in the feed should be marked correctly for eligibility, including is_eligible_search and is_eligible_checkout.

Establishing the technical connection is only the first step in a product feed campaign. The real value comes from sending accurate and current product data through that connection.

Which information is critical in a product feed?

Ad quality in product feed campaigns depends heavily on the fields included in the feed.

Key fields include:

  • Product ID (item_id)

  • Product title (title)

  • Product description (description)

  • Product URL (url)

  • Brand (brand)

  • Category (product_category)

  • Image URL (image_url)

  • Additional image URLs (additional_image_urls)

  • Price (price)

  • Sale price (sale_price)

  • Sale price start date (sale_price_start_date)

  • Sale price end date (sale_price_end_date)

  • Availability (availability)

  • Availability or preorder date (availability_date)

  • Variant group ID (group_id)

  • Variant information (variant_dict)

  • Color (color)

  • Size (size)

  • Size system (size_system)

  • Seller name (seller_name)

  • Seller URL (seller_url)

  • Return policy (return_policy)

  • Returns accepted (accepts_returns)

  • Exchanges accepted (accepts_exchanges)

  • Shipping information (shipping)

  • Search eligibility (is_eligible_search)

  • Checkout eligibility (is_eligible_checkout)

  • Ads eligibility (is_ads_eligible)

  • Ads metadata field (ads_metadata)

Some of these fields are technically required, while others support ad performance and product visibility. In Product Feed campaigns, it is particularly important for title, description, image_url, price, availability, url, and is_ads_eligible to be accurate and up to date.

Some of these fields are technical requirements, while others are important for performance quality. In practice, incomplete or weak product data makes it more difficult for ads to appear with the right message and in the right context.

To explore why product feeds are critical for AI shopping experiences in more detail, read Product Feeds and AI Shopping.

How should product titles and descriptions be written in an OpenAI Product Feed?

In Product Feed campaigns, ad headlines and descriptions are generated from the data in the product feed. Product titles and descriptions therefore become important not only for product pages and SEO, but also for how ads appear.

Weak product title:

“Women’s Dress”

Better product title:

“Strappy Midi-Length Summer Linen Women’s Dress”

Weak description:

“New-season women’s dress.”

Better description:

“A women’s dress suitable for summer events and everyday use, featuring lightweight linen fabric, a midi-length cut, and a strappy design.”

The second example explains the product type, material, use case, and style more clearly, strengthening the advertising experience.

Strategic recommendation: Tools such as Optifeed can automatically generate product information with AI and help advertising systems understand products more effectively.

OpenAI Ads Product Feed Filtering

How are product filters used?

When creating an Ad Group in a Product Feed campaign, product filters can be used to determine which products will be included in that Ad Group.

Product filters can be used to:

  • Select a specific category
  • Include products that are currently in stock
  • Select products within a specific price range
  • Separate discounted products
  • Promote a specific brand or collection
  • Group high-performing products separately
  • Include seasonal products in the campaign

For example, a fashion brand could create the following Ad Group structure:

  • New-season women’s dresses
  • Discounted sneaker models
  • In-stock handbags
  • High-margin accessories

Product filters increase campaign control. Dividing products into meaningful groups rather than placing the entire catalog in a single Ad Group makes performance analysis easier.

What is the purpose of the ads_metadata field?

In the OpenAI Product Feed structure, ads_metadata is an important field that can be used to organize products more strategically in advertising campaigns.

This field allows brands to add labels that are useful for campaign management.

Example ads_metadata values:

  • bidding_tier: high
  • product_line: summer_collection
  • margin_group: high
  • stock_level: strong
  • campaign_priority: seasonal
  • category_focus: dresses

These fields make the Ad Group structure more flexible when standard product filters are not sufficient.

For example, if you want to select products that are not only in a specific category but also have strong availability and high margins, using ads_metadata can be useful.

Strategic recommendation: Do not think of ads_metadata only as a technical labeling field. It is one of the fields through which you can carry campaign strategy into the feed. Labeling products according to margin, availability, season, performance, or campaign priority can make future optimization easier.

Common mistakes in OpenAI Product Feed campaigns

Common mistakes made by e-commerce brands in Product Feed campaigns include:

  • Including the entire catalog without reviewing it first
  • Failing to exclude out-of-stock products
  • Using product titles that are too short or unclear
  • Creating descriptions based only on technical specifications
  • Failing to review image quality
  • Sending prices that do not match the product detail page
  • Sending incomplete or inaccurate variant information
  • Not using the ads_metadata field
  • Grouping too many product categories in the same Ad Group during the first campaign

These mistakes can reduce advertising performance and make campaign management more difficult.

Strategic recommendation: The objective of the first product feed campaign should not be to include the maximum number of products, but to generate reliable learning from the right product set. Scaling should begin only after data quality has been validated.

Example OpenAI Product Feed campaign structure

An e-commerce brand could structure its first OpenAI Product Feed campaign as follows:

Campaign:
USA | Product Feed | New-Season Women’s Fashion | Conversions | July 2026

Ad Group 1:
Summer Women’s Dresses

Product filter:
Category: Women’s Dresses
Availability: In stock
Price: Over USD 200
ads_metadata: seasonal_priority: summer

Ad template:
The primary product image is used.

Review:
The title, description, image, price, and product URL are checked through the sample preview.

This structure provides a simple but manageable starting point. The campaign focuses on one main theme, the product filter is clearly defined, and the ad preview can be reviewed easily.

How should performance be analyzed in Product Feed campaigns?

In Product Feed campaigns, it is not enough to evaluate performance only at the campaign level. Product groups, categories, price ranges, availability, and feed quality should also be considered together.

Questions to review include:

  • Which product groups receive the most impressions?
  • Which products have higher or lower click-through rates?
  • Is the product feed being updated regularly?
  • Which products receive many clicks but generate few conversions?
  • According to GA4 data, how does the conversion rate of traffic from ChatGPT Ads Product Feed campaigns compare with other channels?
  • What characteristics do converting products have in common?
  • Do product groups separated through ads_metadata perform differently?

These analyses help optimize not only the media side of the campaign, but also the underlying product data.

How can Optifeed support this process?

Optifeed is an AI-powered product feed optimization platform that helps e-commerce brands prepare their product data for advertising platforms and AI shopping experiences.

Product data sits at the center of OpenAI Product Feed campaigns. The feed must therefore be created in the correct format, product titles and descriptions should be improved, price and availability should remain up to date, and products should be segmented according to campaign strategy.

Optifeed can support brands with:

  • Preparing Product Feeds for OpenAI Ads
  • Optimizing product titles and descriptions
  • Organizing price, availability, variant, and category data
  • Enriching product attributes and use cases through AI Enrich
  • Segmenting products according to performance, availability, margin, or season
  • Creating campaign labeling structures similar to ads_metadata
  • Managing multichannel feeds for Google, Meta, TikTok, and AI platforms

For a more detailed look at product feed preparation, read How to Prepare an AI Shopping-Ready Product Feed.

About the author
Zafer Kavaklı - Optifeed

Zafer Kavaklı

Co-Founder at Optifeed

Zafer Kavaklı is co-founder of Woom Digital and Optifeed. He is an experienced digital marketer who has been working in the field since 2012. He started his career as a digital marketing intern at Teknosa and then worked at Modanisa as a digital marketing specialist. After that he worked as digital marketing manager at ebebek. Following these roles, he ventured into entrepreneurship by establishing his own performance marketing agency named Woom Digital. Zafer has embarked on a new business venture in the SaaS sector, creating a product management tool named Optifeed.