A product ad template is a creative type in OpenAI's ChatGPT Ads API that builds an ad from your product feed, pulling the image, price, and destination URL from the catalog item ChatGPT shows. A chat card asks you to write and upload the whole creative, while a product ad template fills those details from your feed at serve time.
Selling more than a handful of products forces a choice: hand-build one ad per SKU, or let a single template cover your catalog. Skip it and your ChatGPT Ads coverage caps at the few products you had time to write, missing the shopping conversations where buyers already compare options.
Inside OpenAI's ChatGPT Ads API, a product ad template is the creative object that maps your feed onto a live ad unit, so one definition serves ads across every eligible item in your catalog. You set the creative type to product_ad_template and reference feed fields with macros such as {{product.title}} and {{product.price}}. ChatGPT substitutes the real values when it selects a product.
The template sits at the ad level, below the campaign and the ad group. Because the image and destination come from the matched feed item, it carries no file or target URL of its own the way a chat card does. It needs a feed with the required fields populated and is_ads_eligible set to true, plus an ad group holding the product set it draws from.
A product ad template is one of two creative types in ChatGPT Ads; the other is the chat card, a hand-written unit for one message. Product ad templates belong to product-feed campaigns, which power the e-commerce side of ChatGPT advertising. AirOps tracks them through its AI Ads product, showing where your ads and competitors' ads appear across the prompts buyers type.
Resources: See how to set targeting, creative specs, and policy for ChatGPT ads.
A product ad template turns catalog data into served ads through a fixed sequence, from feed upload to the rendered product.
Upload the feed: Send your product file to OpenAI by CSV or TXT upload, hosted HTTPS URL, or SFTP push, with required fields filled and is_ads_eligible set per item.
Build the template: Create an ad with creative.type set to product_ad_template, using macros like {{product.title}} and {{product.price}} in place of fixed copy.
Attach a product set: Place the template in an ad group that sets which feed items it can serve and the context hints for the conversations you target.
Match and fill: When a conversation fits, ChatGPT selects an eligible item and substitutes its title, body, price, image, and destination.
Render and measure: ChatGPT shows the finished card below its answer, and clicks and conversions report through the Ads API.
The output tells you which products and prompts earn clicks, so you can prune items and sharpen hints. It does not explain why ChatGPT matched a conversation, since targeting runs on conversational context you shape with hints and cannot set as keywords.
Product ad templates decide how much of your catalog can show up in ChatGPT conversations, which shapes whether paid AI search earns a line in your marketing budget. The choice affects coverage, your team's workload, and how fast you can react when prices or stock change.
Catalog-wide coverage: One template can serve every eligible item in your feed, so a large catalog reaches shopping conversations without someone writing a separate ad for every product in the catalog.
Live pricing and stock: Because the creative pulls from your feed, price and availability update when the feed updates, which keeps you from advertising a product that just sold out.
Feed quality becomes the deciding factor: A weak feed is the failure mode here. Missing images, truncated descriptions, or a false is_ads_eligible flag keep products out of the running no matter how strong your bid or budget is.
Growth marketers use a product ad template to put an entire e-commerce catalog into ChatGPT shopping conversations from a single feed-driven creative.
SEO managers use a product ad template to win paid placements for products that competitors already dominate in organic AI answers.
Demand gen leads use a product ad template to test which catalog segments convert in ChatGPT before committing larger budget to the channel.
The product feed is the catalog file you upload to OpenAI, and every field a template renders, including title, body, price, image, destination URL, and the is_ads_eligible flag, comes from the row for the matched item in that uploaded file.
Macros such as {{product.title}} and {{product.price}} are the placeholders you put inside the template so ChatGPT can drop in each product's real values at serve time in place of fixed copy you would otherwise write once.
The product set defined on the ad group controls which feed items a template is allowed to serve, so you can point different templates at different slices of your catalog and keep the context hints tight for each one.
Serve your whole eligible catalog in ChatGPT from one creative definition.
Update prices and availability through the feed instead of editing each ad by hand.
Cut creative production from one ad per product down to a single template.
Reach shopping conversations on ChatGPT where buyers compare products before they decide.
Scale across thousands of SKUs without scaling your ad-ops headcount.
Populate every required feed field (item_id, title, description, url, brand, image_url, availability, and price), because incomplete rows get dropped before they can serve.
Write clear, specific product titles and descriptions, since the template renders exactly what the feed contains and vague copy produces vague ads.
Use a hosted URL or SFTP push instead of one-off manual uploads, so items stay fresh and do not expire out of eligibility.
Keep is_ads_eligible accurate per item, so only products you actually want promoted enter the auction.
Group feed items into tight product sets per ad group, so your context hints match the conversations each slice should serve.
Refresh the feed on a schedule, so the price and stock in your ads track what is true on your site.
Avoid treating the template as set-and-forget. The common mistake competent teams make is uploading a feed once and leaving it, so expired items, stale prices, and sold-out products keep serving until performance quietly erodes.
AirOps: tracks where your product ad templates and competitors' ads appear across the ChatGPT prompts your buyers use, and connects spend, clicks, and CPC to your organic AI search data.
OpenAI Ads Manager and Advertiser API: the native tools where you create the product_ad_template creative, upload product feeds, and manage ChatGPT Ads campaigns.
Shopify: supplies a product feed that can connect into ChatGPT Ads, so store owners can populate the catalog data a template renders.
Audit your feed: This week, before you build anything, pull your existing product feed and check that every required field is present and that is_ads_eligible is set correctly. No budget or approval needed.
Choose a delivery method: Decide between a hosted HTTPS URL or an SFTP push so items refresh automatically and do not expire out of eligibility.
Build one template: Create a single ad with creative.type set to product_ad_template and your feed macros in place, pointed at a small product set to start.
Set context hints: On the ad group, write short context hints that describe the conversations and buyer needs your products fit, since ChatGPT matches on context.
Launch, measure, and expand: Start the campaign paused, review which products and prompts earn impressions and clicks, then widen the product set and refine hints based on what performs.
A product ad template is the ChatGPT Ads creative type that builds ads automatically from your product feed.
You create it by setting creative.type to product_ad_template and mapping feed fields with macros.
The main constraint is feed quality, since only items with complete data and an eligible flag can serve.
The main risk is a stale feed that keeps advertising wrong prices or sold-out products.
The leverage is coverage, because one template can serve your entire eligible catalog across ChatGPT conversations without per-product work.
A product ad template builds the ad from your product feed, while a chat card is a single unit you write and upload yourself. With a chat card, you supply the title, body copy, image, and destination URL, and that one creative runs as written. A product ad template instead references feed fields through macros, so ChatGPT fills in the title, body, price, image, and destination from whichever catalog item it selects for the conversation. That difference shapes how you work: chat cards suit a focused message for one offer or campaign, and product ad templates suit e-commerce catalogs where you want many products eligible at once. The two creative types are the only options in the ChatGPT Ads API today. Product ad templates also belong to product-feed campaigns, so choosing one commits you to maintaining a clean, current feed as the real source of your ad content.
Update the feed as often as your prices, stock, and catalog change, which for most stores means daily or close to it. The feed is the live source for every product ad template, so whatever it says about price and availability is what buyers see in ChatGPT. If you run promotions, adjust inventory through the day, or add and retire products regularly, a daily refresh keeps your ads honest. The safest setup is a hosted HTTPS URL or an automated SFTP push that OpenAI re-reads on a schedule, since manual uploads can lapse and let items expire out of eligibility. There is no benefit to updating more often than your underlying data actually moves, so match the cadence to how fast your data actually moves and no faster. The failure case is predictable: a feed left untouched for weeks will keep serving sold-out items and old prices long after they stop being true.
The same template shows different products because ChatGPT picks which eligible item fits each conversation, and you do not make that pick. A product ad template defines the shape of the ad and the feed fields it renders, but the actual product is chosen at serve time from the product set on your ad group. ChatGPT weighs the context of the conversation and your context hints, then selects the catalog item it judges most relevant and substitutes that product's details into the template. Because conversations differ, the matched product differs too, and the same template can surface a running shoe in one chat and a hiking boot in another. This variation is the point of the format: one definition adapts across many buyer questions. It also means your influence is indirect. You shape outcomes by curating which items are eligible, writing precise hints, and keeping feed data complete, then letting the match happen.
No, you cannot force a specific product into a specific conversation, and trying to is the wrong mental model for this format. A product ad template hands product selection to ChatGPT, which matches an eligible item from your product set to the conversation in front of it. What you do control is the pool and the signals. You decide which items carry is_ads_eligible, which products sit in each ad group's product set, and what your context hints say about the conversations you want to reach. If you need tight control over a single message or offer, a chat card is the better instrument, since it runs exactly the creative you write. For catalog advertising, the practical move is to narrow the product set and tighten hints per ad group, so that even though ChatGPT makes the final pick, every product it can choose is one you are happy to show.
A good setup is one where feed data is complete and current, product sets are tightly scoped, and early performance shows real clicks and conversions on the products you care about. Before you raise budget, confirm that every required field is populated, that is_ads_eligible is true only for products you actually want promoted, and that your delivery method refreshes automatically. Launch paused, then watch which products and prompts earn impressions and clicks over a few days of real data instead of one snapshot. Strong signals look like steady impressions on relevant conversations, click-through that holds up, and conversions that justify the cost per click you are paying. Weak signals, such as impressions on off-topic prompts or clicks that never convert, usually trace back to loose hints or an overbroad product set. Fix those at the ad-group level first, then expand the product set and budget once the small version is working.