ChatGPT Ads Manager opens product feeds - what Amazon brands do now

What's happening

On June 2, 2026 OpenAI moved product feed management into a dedicated Feeds section inside the ChatGPT Ads Manager. Retailers connect the same structured catalog they already send to Google Shopping; the platform then auto-generates sponsored placements from product names, images, and attributes. Coverage was confirmed by PPC Land, Digiday, and eMarketer.

The mechanic is closer to Google Shopping or Performance Max than to Amazon Sponsored Products. Advertisers do not pick keywords or match types. The platform ingests the catalog, builds ad units automatically, and serves them below ChatGPT responses based on conversational intent. The system supports up to one million SKUs per advertiser with a 100-product sample required before opening up the full catalog.

Why it matters for Amazon brands

ChatGPT is the second large surface after Rufus where shoppers are guided by conversational intent rather than keyword search. For Amazon-first brands, the question is not whether to abandon Sponsored Products. The question is whether the catalog signal you already maintain for Google Shopping is clean enough to spin up a learning channel without a major operational lift.

Brands tracking PPC analytics at the product level on Amazon can now extend the same discipline to ChatGPT by reusing one feed across two surfaces. The conversational placement does not behave like a one-for-one substitute for keyword-targeted Sponsored Products, which makes incremental ROAS comparison the right early metric, not blended ROAS.

What you should do now

1. 1.

   Audit your Google Shopping feed for ChatGPT reuse

   Product titles, attributes, and primary images now power both organic ChatGPT product discovery and paid placements through the same file. Make titles entity-clear, fill brand and category attributes, and use lifestyle images that read at thumbnail size.

2. 2.

   Submit the 100-product sample for high-revenue SKUs first

   Do not lead with long-tail or seasonal items. Pick the 100 SKUs that already convert hardest on Amazon and Google so the platform's learning phase optimizes against your best performers, not your weakest.

3. 3.

   Instrument conversions before June 5

   If you want to qualify for the June 5 conversion-optimized rollout, OpenAI requires at least one conversion event flowing by June 1 via the JavaScript Pixel or Conversions API. Miss that window and you sit on impression-driven pacing for the first weeks.

4. 4.

   Compare incremental ROAS against Amazon Sponsored Products

   Start with a small slice of always-on budget. Track contribution margin per channel in your P&L and compare ChatGPT incremental ROAS against Amazon Sponsored Products in the same category before reallocating budget. Conversational intent is a different signal; treat it as a learning channel for at least two cycles.

How Nova helps

Nova reconciles product-level ad spend against retailer fees and unit margin in one P&L view. When a new channel like ChatGPT enters the mix, agencies and brand managers can isolate Amazon Sponsored Products contribution by SKU and compare it against blended off-Amazon results without losing the underlying Amazon view.