On this page
The single-image Sponsored card is the format everyone writes about. Product-feed ads are the ones that matter more if you sell physical products, and they work on completely different inputs.
You do not design a creative for them. Ads Manager generates them from your product catalogue: product name, image, attributes and price. That means your feed is the ad, and feed hygiene stops being a back-office chore and becomes creative work.
How we got here, briefly. A Shopify-powered product-spotlight carousel was reported from March 2026. Product feed campaigns appeared in Ads Manager in May 2026, tested with ad partners (Shopifreaks, reported). OpenAI's ChatGPT Ads app for Shopify opened to US merchants on 16 September and was scheduled to reach other ChatGPT Ads markets from 23 September (Novoads, reported; OpenAI's help centre still said "shortly" that morning). So Shopify is now one way to sync a catalogue, not the definition of the unit.
Last verified: 2026-11-01. Every dated change to the channel is in our ChatGPT Ads guide. LocalAds makes product imagery and does not place ads or manage feeds. Confirm current requirements inside Ads Manager before you build.
What changed in September and October
Three changes, all reported by Common Thread Collective rather than confirmed by OpenAI directly:
- Bulk product campaigns (30 September). Under Campaigns, then Bulk uploads, you can download a product feed template, fill it in and upload it. Ads Manager generates multiple campaigns, ad groups and ad templates from the one sheet.
- Per-product review status. The Products tab now shows where each product sits in review and why rejected items failed. A rejected product does not appear in shopping results or ads.
- Virtual try-on (1 October). Clothing and accessories listings can carry a Try on button that shows the shopper wearing the item from a selfie. Only brands with active product feed campaigns appear in these moments.
The common thread: every one of these runs off the feed. Bulk campaigns are built from it, review happens per feed item, and try-on is gated on having a feed campaign live. Our reading, not a published rule: try-on works from your listing image, so a clean, unobstructed product shot matters even more for apparel than it did before.
Feed as creative
With the single-image card you control everything: you choose the square, you write the title, you pick the landing page.
With product-feed ads you control almost nothing at ad time. The unit assembles from what your catalogue already says: the product image, the title, the live price, the availability. If your feed image is a grey-background packshot with a watermark and your title is Clutch Bag Red PU Leather Party Wear Ladies Handbag 2026 New, that is your ad. There is no creative layer to rescue it.
So the work moves upstream, and it is unglamorous: clean images, honest titles, accurate prices, correct stock states.
What the images need to be
Square, clean, product dominant, consistent across the catalogue. Same principles as the Sponsored card image, with one addition: consistency matters more here because your products appear in a carousel next to each other and next to competitors.
A studio treatment on a plain background is the safest default:
Product isolated, even lighting, neutral ground, nothing cropped. This survives being shrunk into a carousel cell and sits comfortably beside other products.
Lifestyle can work for the hero image, if the product still dominates:
Context without losing the product. The risk with lifestyle in a feed is inconsistency: if half your catalogue is styled and half is packshots, the carousel looks assembled from two different stores.
What fails, specifically:
- Watermarks and retailer logos. They read as scraped content.
- Burnt-in price flashes or sale badges. The unit shows live price already, and a stale "40% OFF" baked into the image contradicts it.
- Collage or multi-pack images where five variants share one frame. In a small cell nothing is identifiable.
- Inconsistent backgrounds across the catalogue, which makes the carousel look untrustworthy even when each image is individually fine.
- Text overlays of any kind. Same reason as the Sponsored card: the unit supplies its own text furniture.
The feed fields that decide whether you show
Title. Write it for a person, not a search engine. Lovers Clutch, quilted leather, red beats a keyword-stuffed string. Front-load the thing the product actually is.
Price. Must be live and correct. Feed price disagreeing with landing page price is the most common cause of products being suppressed across every shopping surface, and there is no reason to expect this one to be more forgiving.
Availability. Out-of-stock products in an active feed waste impressions and produce the worst possible click.
Image URL. Must be reachable, high resolution, and stable. Images behind a CDN that rate-limits or expires are a recurring failure.
Product identifiers and attributes. Populate them properly. Sparse attributes limit how confidently the system can match your product to a conversation.
Review status. Check the Products tab after every upload. A product rejected at review is simply absent, and the reason is listed there rather than emailed to you.
None of this is exotic. It is the same discipline that governs Google Shopping and Amazon listings, which is genuinely good news: work you have already done for those surfaces largely carries over.
The overlap with Amazon listing images
If you sell on Amazon you have already solved a version of this problem: a set of clean, square, conversion-oriented images per product, consistent across the catalogue, no watermarks, no stale badges.
The requirements are not identical, but the discipline is the same and the assets are often reusable with light rework. If you are starting from nothing, our Amazon listing images guide covers how to build a clean nine-image set from a product URL, and most of that set is directly usable as feed imagery.
A sensible order of work
- Fix price and availability accuracy first. It is the cheapest fix and the most common disqualifier. Then clear anything flagged in the Products tab.
- Rewrite titles for humans, starting with your top 20 products by revenue.
- Standardise on one image treatment across the catalogue. Pick studio or styled and stick to it.
- Replace the worst images, prioritising by revenue rather than by how bad they are.
- Only then worry about the single-image Sponsored card creative, which is a separate build covered in how to create ads for ChatGPT.
That order matters because feed problems suppress products silently, while creative problems merely underperform visibly.
Where LocalAds fits
LocalAds generates square product imagery from your real product page, so packaging, colours, and details stay accurate rather than being invented, and the treatment stays consistent across a catalogue. Both images above are unedited output.
We do not manage your feed, submit it, place ads, or guarantee that any product appears. We make the images the feed points at.
Changelog
- 2026-10-13: Reframed from a Shopify-only carousel to product feed campaigns. Added bulk product campaigns, per-product review status and virtual try-on (all reported), and the ChatGPT Ads app for Shopify reaching markets beyond the US.
- 2026-08-17: First published.
FAQ
What are ChatGPT shopping ads? Ads generated from your product catalogue, run as product feed campaigns in ChatGPT Ads Manager. Unlike the single-image Sponsored card, you do not design a creative; the ad is built from your product names, images, attributes and price, so your catalogue data and imagery are the ad. Since 30 September you can also create these campaigns in bulk from a feed template (reported).
What image size do ChatGPT shopping ads use? Square imagery, consistent with the rest of the format. Aim for clean, high-resolution square product images with the product dominant, no watermarks, no burnt-in badges, and one consistent treatment across the catalogue. Confirm current requirements inside Ads Manager before building.
Do I need a Shopify store for ChatGPT shopping ads? No. The ChatGPT Ads app for Shopify is one way to sync a catalogue (US from 16 September 2026, other markets scheduled from 23 September, reported). Ads Manager also accepts a product feed uploaded from its own template under Bulk uploads. If you have no feed at all, the single-image Sponsored card is still available, and the same square creative discipline applies.
Why is my product not showing in ChatGPT shopping ads? Check the Products tab first: it shows review status and the rejection reason per product. Beyond that, the usual suspects are feed problems rather than creative ones: price mismatch between feed and landing page, stale availability, unreachable or low-resolution image URLs, and sparse product attributes. Fix data accuracy before you touch imagery.
Can I reuse my Amazon listing images for a ChatGPT product feed? Often yes, with light rework. Both surfaces want clean, square, consistent product imagery without watermarks or stale promotional badges. Amazon sets that include heavy infographic text are less transferable than the clean hero and in-context shots.
What is ChatGPT virtual try-on? A Try on button on eligible clothing and accessories listings that shows the shopper wearing the item, launched 1 October 2026 (reported). Only brands with active product feed campaigns appear in those moments, which makes a live, clean feed a requirement rather than a nice-to-have for apparel brands.
Should I do feed work or Sponsored card creative first? Feed work. Feed problems suppress products silently, while weak creative merely underperforms in a way you can see and fix.
The takeaway
Product-feed ads remove the creative layer and expose your catalogue directly. Bulk campaigns and try-on make the feed matter more, not less. That is an advantage if your feed is clean and a liability if it is not. Get price, availability, titles, and image consistency right first, because no amount of good creative elsewhere compensates for a product that never surfaces.
Related reading: