ChatGPT Product Feed Ads: Everything Retailers Need to Know in 2026

Henry Purchase

Co-Founder

In May 2026, OpenAI rolled out product feed ads inside ChatGPT. Retailers can now upload their full product catalogue, set a few rules, and let ChatGPT generate ads automatically based on what users are asking about.

For most e-commerce brands, this is the single most important update since the ad platform launched. Running ads at scale was the bottleneck. Product feed ads remove it.

This post walks through what they are, how they work, who they're built for, and how to launch your first feed-based campaign.

Want the Focal team to set up your product feed and run the campaigns for you? Book a free 30-minute discovery call. We've been shipping ChatGPT Ads campaigns for retailers since the channel opened.

TL;DR: What You Need to Know

What Are ChatGPT Product Feed Ads?

ChatGPT product feed ads are sponsored placements inside ChatGPT generated automatically from a retailer's product catalogue. Instead of writing ad copy product by product, the retailer uploads a structured feed (titles, descriptions, images, prices, availability), and OpenAI's system creates and serves the ads.

The ads themselves look the same to ChatGPT users. They appear beneath responses, clearly labelled as sponsored. What's different is the back end. Each ad is built from live product data, so titles, descriptions, images, and links are always current.

If you've ever run Google Shopping or Meta Catalog ads, the model will feel familiar. You're handing over the catalogue and letting the platform match products to high-intent moments. The difference is what counts as a high-intent moment. On Google, it's a search query. Inside ChatGPT, it's a conversation about a need, a comparison, or a decision.

How They Work: The 7-Step Process

OpenAI documented the full setup process in their Ads Manager help docs. Here's the simplified version.

Step 1. Create the feed inside Ads Manager

In Ads Manager, go to the Tools tab and select Feeds. Click Create Feed and follow the prompts. Once the feed is created, Ads Manager generates the SFTP connection details you need to upload data.

Step 2. Configure your SFTP connection

From the feed row, open the three-dot menu and select Edit SFTP Connection. Pick your authentication method (password or SSH key), then follow the in-app connection guide.

This is the most technical part of the setup. If your team hasn't worked with SFTP before, set aside a couple of hours and a dev resource. Most of the support tickets in the beta have been here. Focal's team handles SFTP setup for clients as part of onboarding.

Step 3. Upload and validate

Upload the feed through the SFTP connection. Wait for Ads Manager to process it. Small feeds clear in minutes. Larger feeds (closer to the 2 million product ceiling) can take a few hours.

Items process gradually. Check the ad group creation modal to see how many items are ready to serve. Products that don't pass validation get flagged with a reason, so you can fix and re-upload.

Step 4. Create the campaign

In the Campaigns tab, create a new campaign and select "Product feed" from the campaign type dropdown. Complete the standard campaign settings (budget, schedule, geo, targeting).

Step 5. Build the ad group and filter products

Create an ad group for the campaign and select the product feed you want it to run from. Then apply filters to define which products are eligible.

The filters in the UI cover the obvious cuts (brand, category, price band, stock status). If you need more granular control, OpenAI lets you add custom values via the ads_metadata field in your feed. So you can create your own segments like bidding_tier: premium or product_line: summer_2026, then use them to define ad groups.

Review the product count shown in Ads Manager to confirm how many products pass your filters. If it's lower than expected, the filters are too tight or some products failed validation.

Step 6. Create the ad template

Each ad group needs one ad template. The template defines the image field. Titles and descriptions come from the feed directly. Currently you can only set the image, so creative testing happens at the catalogue level (different lifestyle images per product), not at the campaign level.

Use the sample product preview to confirm titles and descriptions render the way you want.

Step 7. Review and launch

Before submitting, confirm:

• The correct feed is selected
• The product count matches the intended set
• Titles, descriptions, images, and landing pages all preview correctly
• Campaign, ad group, budget, bid, and targeting settings are complete

Submit for review and launch.

Why ChatGPT Product Feed Ads Matter Right Now

Three reasons this is a bigger update than it looks.

1. The bottleneck is gone

The biggest barrier to running ChatGPT Ads at scale was that every campaign had to be set up manually. For a brand with 50 SKUs, that's doable. For a retailer with 50,000, it wasn't. Product feed ads turn the catalogue into the campaign. The work moves from "build campaigns" to "structure your feed and set the rules."

2. Performance signals are strong

OpenAI has said publicly that ads built from product feeds have been "among the strongest-performing ads" on the platform in the beta. Take that with appropriate calibration (OpenAI has an interest in saying so), but the structural reason holds. Feed ads run on accurate, current product data and serve against high-intent conversations. The match quality is naturally higher than running a static text ad against the same query.

3. Organic and paid product visibility are converging

In Search Engine Land's coverage of the launch, Anu Adegbola noted that previously, product data could inform ChatGPT's organic answers but couldn't power advertising. That changes with feed ads. The same structured catalogue can drive both organic mentions and paid placements, which is how shopping campaigns work on Google.

For retailers, this means the work of getting product data clean (titles, descriptions, attributes, images) compounds across both organic and paid surfaces. The cost of doing it well dropped.

Who ChatGPT Product Feed Ads Are Built For

Three categories where feed ads make sense in 2026:

Mid-sized to large retailers (5,000+ SKUs). This is the obvious fit. If you have a catalogue this size, manual ad creation never scaled and you've been waiting for a feed-based option. Feed ads are how you get into the channel at all.

Multi-brand retailers. If you sell other brands' products (Amazon-style, marketplaces, multi-brand DTC), feed ads let you test which brands resonate with ChatGPT's user base before you double down. The catalogue is the test.

Retailers already running Google Shopping. Your existing feed is most of the work. The format is similar, the structure transfers, and the operational muscle (feed hygiene, attribute completeness, image quality) already exists. The lift to add ChatGPT product feed ads to an existing Google Shopping operation is small. Focal's Google Ads connector maps existing Shopping feed data into the OpenAI feed specification automatically.

Feed ads don't make sense for everyone:

• Below 1,000 SKUs. That's the platform minimum. If you're under it, run standard ChatGPT Ads campaigns.
• Highly considered, low-volume products. A $50,000 enterprise software contract or a $200,000 industrial machine isn't a product feed ad use case. The decision happens over multiple conversations, not one impulse moment.
• Categories where ChatGPT Ads aren't open yet. Some categories are still in restricted rollout. Book a discovery call and the Focal team will check if your category is live.

What Performance Has Looked Like So Far

A few patterns are worth flagging based on the campaigns we've seen across the beta:

Catalogue completeness drives performance harder than bid. Brands that wrote one-line product descriptions and tested cheap stock images underperformed brands that filled in every feed attribute (brand, gender, age group, colour, material, lifestyle image). Feed ads reward the catalogue work, not the bidding work.

Conversational intent matches differently than search intent. A ChatGPT user asking "what's a good gift for my dad who likes hiking" surfaces a different product mix than a Google search for "men's hiking gifts." The conversation has more context: budget hints, occasion, recipient. Brands whose catalogue depth covers context (occasion, age group, use case) get matched more often.

Re-uploads matter. Feeds aren't fire-and-forget. Stock changes, prices change, new products launch. The brands that re-upload daily (or schedule automatic feed refreshes) compound the early advantage. The ones uploading monthly slide as their feed drifts out of sync.

Most retailers are integrating Google or Meta data alongside. Existing audience signals, conversion data, and creative performance from Google Shopping or Meta Catalog campaigns predict which products will perform on ChatGPT too. The Focal Google Ads and Meta Ads connectors are how we pipe that data in.

ChatGPT Product Feed Ads vs Google Shopping

If you're already running Google Shopping, ChatGPT product feed ads will feel familiar. Here's where they overlap and where they diverge.

The strategic move for most retailers in 2026 isn't to choose between the two. It's to run both and let the comparison data tell you which channel is converting at lower CAC for your category. Most of the retailers we work with are running both side by side.

For a deeper breakdown of how the two channels compare, see our ChatGPT Ads vs Google Ads guide.

Common Mistakes to Avoid

Five things to get right before you launch.

1. Don't upload a Google Shopping feed directly

OpenAI's feed spec is different from Google Merchant Center's. The attribute names, required fields, and image dimensions vary. Map your existing feed to OpenAI's spec before uploading or the validation step will reject most products. The Focal Google Ads connector handles this mapping automatically.

2. Don't skip the ads_metadata field

The UI filters work for basic cuts, but the real targeting power is in ads_metadata. Use it to add custom values like bidding tier, product line, margin segment, or seasonal tag. Then build ad groups around those segments. Without this, you'll either over-broaden (one ad group for everything) or be forced to launch dozens of identical campaigns by category.

3. Don't ship one ad group for the whole catalogue

This is the most common beta mistake. Splitting by margin (high-margin products in a high-bid ad group, low-margin in a lower-bid group) is the difference between feed ads being profitable and feed ads burning budget evenly across stock you can't afford to discount.

4. Don't set and forget

Re-upload your feed daily if possible. Stock changes, prices change, new products launch, old products go out of stock. Stale feed data is the fastest way to waste spend on products that aren't available. Focal's alerts and reports surface feed drift the day it happens.

5. Don't try to creative test inside ChatGPT Ads Manager

Currently the ad template only lets you control the image field. Titles and descriptions come from the feed. If you want to creative test, do it at the feed level (different lifestyle images, different product description angles), not the campaign level. The brands that figure this out early get a measurable edge.

How Focal Helps Retailers Run Feed Ads

Focal is the done-for-you ChatGPT Ads agency that runs the full feed-based campaign workflow for retailers. The work breaks into four parts:

Feed audit and mapping. We audit your existing product data (Google Shopping feed, PIM exports, or raw catalogue), then map it into OpenAI's feed specification with the attributes that drive performance. This is the foundation. Brands that skip the mapping work cap their performance from day one.

SFTP setup and feed automation. We configure the SFTP connection and schedule automated daily uploads, so your feed stays in sync with stock, prices, and new launches without your team touching it.

Campaign structure and ad group strategy. We build the ad group strategy from your margin and inventory data, not from generic best practices. Different products earn different bids. We structure that into the campaign from launch using the Bulk Ad Launcher.

Ongoing optimisation and reporting. Daily campaign management, weekly creative refreshes (at the feed level), monthly performance reporting tied to revenue. We also surface what your competitors are running in our ChatGPT Ad Library so you can adjust before they pull ahead.

If you want a specialist team running feed-based campaigns end to end, book a 30-minute discovery call. The call is free and we'll tell you straight whether your catalogue is ready for ChatGPT product feed ads.

Three other ways to get started without booking a call:

• Try the free ChatGPT Ads tools. No signup, no email gate.
• Join the FutureProof community. 500+ operators sharing what's working on ChatGPT Ads in real time.
• Join the platform waitlist. 1,500+ businesses already on it.

Frequently Asked Questions

What are ChatGPT product feed ads?

ChatGPT product feed ads are sponsored ads inside ChatGPT generated automatically from a retailer's product catalogue. Retailers upload a structured feed of titles, descriptions, images, and prices through OpenAI's Ads Manager, and the system creates ads from that data and matches them to high-intent conversations.

How do I set up a ChatGPT product feed ad campaign?

Create a feed inside OpenAI's Ads Manager, configure an SFTP connection, upload your product data in OpenAI's feed specification, then create a campaign with the "Product feed" type. Build an ad group with filters that match the products you want to run, set the ad template image, and launch. The full setup is documented in OpenAI's Ads Manager help center. Focal handles all of this for clients if you'd rather not set it up internally.

How many products do I need to run feed ads?

The minimum feed size is 1,000 products. The maximum is 2 million. Retailers below 1,000 products should run standard ChatGPT Ads campaigns instead.

Are ChatGPT product feed ads better than Google Shopping?

Different, not necessarily better. ChatGPT product feed ads serve against conversational intent (a user asking ChatGPT for a recommendation). Google Shopping serves against search intent (a user searching a query). For most retailers in 2026, the right answer is to run both and let the data show which channel converts at lower CAC. The two are complementary, not competitive.

How much do ChatGPT product feed ads cost?

ChatGPT Ads uses a CPC (cost-per-click) bidding model, with CPA models reportedly in development. You set your own bids. The total cost depends on your budget, category competition, and bid strategy. There's no platform minimum spend. The minimum useful spend (the level at which you have enough data to optimise) is typically $5,000 to $10,000 per month for a retailer with a meaningful catalogue.

Can Focal manage my ChatGPT product feed ads?

Yes. Focal is a done-for-you ChatGPT Ads agency that runs feed-based campaigns for retailers end to end. We handle feed mapping, SFTP setup, campaign structure, ad group strategy, and ongoing optimisation. Book a discovery call to discuss.

Do ChatGPT product feed ads work for service businesses?

No. The product feed format is built for retail catalogues with discrete products. Service businesses should run standard ChatGPT Ads campaigns instead. Focal runs both.