Preparing a product feed for ChatGPT Ads is not simply a technical file creation process. Product titles, descriptions, prices, availability, images, variants, and category information within the feed affect how ads appear, which products are included in campaigns, and how quickly users can understand each product.
In OpenAI Product Feed campaigns, products are transformed into ads directly from the catalog. The product feed therefore becomes the campaign’s primary data source.
If you need a general introduction to OpenAI advertising, you can first read What Are OpenAI Ads? What Do ChatGPT Ads Mean for Brands?
In this article, we will focus specifically on the areas e-commerce brands should prioritize when preparing product feeds for ChatGPT Ads.
Why is a product feed important for ChatGPT Ads?
A product feed is a structured data file that allows you to transfer products to OpenAI Ads Manager at regular intervals. In OpenAI Product Feed campaigns, the feed carries fields such as product title, description, image, price, availability, and landing page information into the advertisement.
For this reason, feed quality directly affects ad quality.
The difference between a Product Feed campaign and product feed preparation
An OpenAI Product Feed campaign is the process of creating catalog-based ads within Ads Manager by using a product feed. Product feed preparation is the process of making the product data that will be used in the campaign accurate, current, and suitable for advertising.
| Concept | Description |
|---|---|
| Product Feed campaign | The advertising setup process. |
| Product feed preparation | The product data preparation process. |
To explore the campaign structure in more detail, read How Do OpenAI Product Feed Campaigns Work? A Guide for E-commerce Brands.
The focus of this article is how the feed should be prepared before launching a Product Feed campaign.
The first feed preparation check: Required fields
According to the OpenAI product feed specification, certain fields are critical for products to be processed correctly. When these fields are missing, inaccurate, or submitted in a non-standard format, using products in advertising may become more difficult, and the product feed may not be uploaded successfully to OpenAI Ads Manager.
Key fields include:
| Field | Technical field name | Why is it important? |
|---|---|---|
| Product ID | item_id |
Identifies the product uniquely and consistently. |
| Product title | title |
Helps users understand the product in the ad. |
| Product description | description |
Explains the product’s features and use cases. |
| Product URL | url |
Directs the user to the correct product page. |
| Brand | brand |
Supports product identity and trust. |
| Image URL | image_url |
Allows the product to be represented visually in the ad. |
| Price | price |
Influences the user’s purchase decision. |
| Availability | availability |
Shows whether the product can currently be purchased. |
| Ads eligibility | is_ads_eligible |
Determines whether the product can be processed for ChatGPT Ads. |
For a more technical field list and detailed OpenAI product feed specifications, read OpenAI Feed Specifications: Product Feed Fields, File Upload, and API Structure.
How should the product title be prepared?
The product title is one of the most critical fields in the feed because the ad headline in Product Feed campaigns is generated from product feed data.
| Title type | Example |
|---|---|
| Weak product title | Women’s Dress |
| Better product title | Strappy Midi-Length Summer Linen Women’s Dress |
Although the ideal structure varies by category, a strong product title may include:
- Product type
- Brand
- Material
- Color
- Size or dimensions
- Use case
- Distinctive feature
However, the title should not become an unnecessary collection of keywords. It should explain clearly what the product is.
Strategic recommendation: When preparing product titles for ChatGPT Ads, consider not only SEO keywords but also the information users need when making a decision. The title should clearly answer the questions “What are we selling?” and “Who is this product for?” in a single phrase.
How should the product description be written?
The product description should include not only technical specifications, but also the product’s use cases, differentiating features, benefits, and answers to the main questions users may have during the purchase decision.
| Description type | Example |
|---|---|
| Weak description | New-season linen dress. High-quality fabric. Stylish design. |
| Better description | A women’s dress suitable for summer events, holiday outfits, and everyday use, featuring lightweight linen fabric, a midi-length cut, and a strappy design. |
| Much better description | Made from lightweight and breathable linen fabric, this strappy midi-length women’s dress provides comfortable wear in warm weather. Its midi-length cut makes it suitable for everyday city outfits as well as summer events, holiday evenings, graduation ceremonies, and weekend gatherings. Its key differentiators are the lightweight fabric, minimalist strappy design, and the ease with which it can be styled with different shoes, bags, and accessories. It can be paired with sandals for a casual look or with heels for a more elegant outfit. Frequently asked questions: Which season is this dress suitable for? It is suitable for everyday wear, holidays, and events during spring and summer. Is the fabric comfortable? Its linen texture provides lightweight and breathable wear. Where can it be worn? It is suitable for everyday use, summer events, holiday evenings, graduation ceremonies, and weekend gatherings. How can it be styled? Pair it with sandals, a straw bag, and minimal jewelry for a casual look, or with heels and elegant accessories for a more sophisticated outfit. |
If the description is too short, the product may not be understood sufficiently. If it is too long and repetitive, the advertising experience may become weaker. The description should therefore balance product features, use cases, differentiating points, and the key questions that influence the purchase decision.
Optifeed’s AI Enrich feature can generate entirely new product titles and descriptions based on existing product information. This helps both users and AI tools such as ChatGPT understand products more effectively. More descriptive titles, product descriptions that explain use cases, and information that supports the decision-making process can increase product visibility within AI-powered shopping experiences.
Price and availability information should be current
When preparing a product feed for ChatGPT Ads, price and availability information must remain up to date.
One of the most common issues in e-commerce is a mismatch between the price in the feed and the price displayed on the product page. This can reduce user trust and negatively affect advertising performance.
| Field | Technical field name | Check |
|---|---|---|
| Price | price |
Should be submitted correctly together with the currency code. |
| Sale price | sale_price |
Should be lower than or equal to the standard price. |
| Sale start date | sale_price_start_date |
Should be consistent with the campaign period. |
| Sale end date | sale_price_end_date |
Should contain a current and accurate date. |
| Availability | availability |
Should use supported values such as in_stock, out_of_stock, pre_order, or backorder. |
| Preorder date | availability_date |
Should contain the correct date for preorder products. |
Strategic recommendation: Do not include products with recurring availability problems in your initial campaigns. The most suitable test group for ChatGPT Ads consists of products with reliable availability, current pricing, and fully functional landing pages.
Image quality affects advertising performance
The primary product image is submitted through the image_url field. The image URL should be accessible, the image should accurately represent the product, and it should appear clean in the ad preview.
Basic checks for a strong product image include:
- The product should be clearly visible.
- The image URL should work correctly.
- The primary image should display the correct product variant.
- Low-resolution images should not be used.
- The primary image should not contain excessive text or a distracting background.
- Additional images should show the product from different angles or provide useful information about its use cases.
Category and brand information should be accurate
Category and brand fields are important for grouping products correctly and using them more effectively within campaign structures.
In large catalogs especially, a weak category structure makes it difficult to divide products into the right Ad Groups.
| Category type | Example |
|---|---|
| Weak category | Dresses |
| Better category | Women’s Clothing > Dresses > Summer Dresses |
The category structure can be used for:
- Product filtering
- Campaign grouping
- Ad Group segmentation
- Performance analysis
- Seasonal product selection
- Product segmentation
The brand field is also especially important for marketplaces, multi-brand retailers, and category-based advertising campaigns.
Variant information should be complete
Many e-commerce products have variants such as color, size, dimensions, material, or model. When variant information is missing, the correct product option cannot be recommended to the user, and product visibility in ads may decrease.
| Field | Technical field name | Use |
|---|---|---|
| Variant group ID | group_id |
Groups variants belonging to the same product family. |
| Variant information | variant_dict |
Carries variant values such as color, size, and material. |
| Color | color |
Color information. |
| Size | size |
Size information. |
| Size system | size_system |
Size system such as US, EU, or TR. |
| Gender | gender |
Target gender information. |
| Group product title | item_group_title |
The shared title of the variant family. |
Variant data directly affects campaign quality, especially in fashion, footwear, accessories, furniture, and electronics categories.
How should the ads_metadata field be used?
ads_metadata is a flexible field that can be used to label products according to advertising strategy. It is not required, but it can be useful for organizing products more effectively in Product Feed campaigns.
| Metadata field | Example value | Use case |
|---|---|---|
bidding_tier |
high |
Separating high-priority products. |
product_line |
summer_collection |
Creating collection-based campaigns. |
margin_group |
high_margin |
Selecting high-margin products. |
stock_level |
strong |
Separating products with reliable availability. |
campaign_priority |
seasonal |
Labeling seasonal products. |
performance_group |
best_seller |
Grouping best-selling products. |
This field can make the campaign structure more flexible when standard product filters are not sufficient.
Strategic recommendation: The
ads_metadatastructure should not be defined by the technical team alone. Performance marketing, e-commerce, and product teams should work together to determine which product groups should receive priority in advertising.
Eligibility fields should be reviewed
The OpenAI product feed contains eligibility fields that determine which experiences a product can be used in.
| Field | Meaning |
|---|---|
is_eligible_search |
Indicates whether the product can appear in ChatGPT Search. |
is_eligible_checkout |
Indicates whether the product is eligible for checkout within ChatGPT. |
is_ads_eligible |
Indicates whether the product can appear in ChatGPT Ads. |
The is_ads_eligible field is particularly important for ChatGPT Ads. Products that should not be used in advertisements should not be marked as true accidentally. Similarly, this field should be configured correctly for products that should be included in advertising.
The landing page and feed should be consistent
The information in the product feed should be consistent with the information on the product page. After clicking an ad, users should not encounter a different price, different availability status, or incomplete product information.
Areas to review include:
- Does the product title in the feed match the product shown on the page?
- Is the feed price the same as the landing page price?
- Is the product shown as available?
- Can users select the available variants on the page?
- Does the image represent the same product as the feed?
- Does the product URL lead to the correct page?
- Does the page load quickly on mobile devices?
- Are return, delivery, and trust-related details clearly visible?
Landing page consistency is important not only for user experience, but also for ad quality and conversion rate.
Recommended workflow for product feed preparation
You can adapt the following workflow when preparing a product feed for ChatGPT Ads:
- Prepare the product catalog.
- Identify products that should not be used in advertising.
- Check whether the required OpenAI feed fields are present in the product feed.
- Make product titles clear and descriptive.
- Strengthen product descriptions with use cases and decision criteria.
- Update price, availability, and sale fields.
- Check that image URLs work correctly.
- Organize category and brand data.
- Submit complete variant information.
- Mark campaign priorities using
ads_metadata. - Compare landing page information with feed data.
- Validate the structure using a test feed or sample products.
- Select a controlled product set for the first campaign.
This workflow does more than make the feed technically valid. It also creates a stronger foundation for advertising performance.
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.
When preparing product feeds for ChatGPT Ads, Optifeed can help with:
- Improving product titles and descriptions
- Keeping price, availability, and variant data up to date
- Organizing category, brand, and product attribute data
- Enriching key product features and use cases through AI Enrich
- Segmenting products according to performance, availability, season, margin, or campaign priority
- Creating advertising labeling structures similar to
ads_metadata - Managing multichannel feeds for platforms such as Google, Meta, TikTok, and OpenAI
The objective is not simply to prepare a separate feed file for OpenAI. Optifeed’s approach transforms product data into a clean, current, and meaningful structure that can be used across advertising and AI shopping channels.
