Enterprise

ChatGPT Ads opens customer-list targeting as LiveRamp joins OpenAI's ad stack

A new LiveRamp partnership and OpenAI's own Ads Manager uploader turn ChatGPT into a list-targetable channel — but a 25,000-match floor sets the bar for who can run inclusion campaigns today.

Photo: Unsplash / Carlos Muza — Laptop screen showing marketing analytics dashboard with charts and graphs

LiveRamp announced on September 28 that it has become a technology partner of ChatGPT Ads, extending its RampID identifier into OpenAI’s Custom Audiences across 11 of the 63 markets where OpenAI currently sells inventory. In practical terms, ChatGPT just became another destination you can point a customer list at.

The LiveRamp route is designed for advertisers already running first-party data through an identity graph, the kind of operation that would treat ChatGPT as another destination in a managed first-party data program. But OpenAI Ads Manager also accepts direct uploads. According to its own documentation, advertisers can create Custom Audiences by dropping in a CSV or TXT file of email addresses, phone numbers, hashed variants, or Google Advertising IDs. No middleware required.

The gating mechanism is a number: 25,000.

That’s the matched-user floor OpenAI’s July documentation set for inclusion campaigns, with 100,000 recommended. A product update the week of August 31 removed the floor for exclusion audiences and added Google Advertising ID as an accepted identifier, but left the 25,000 inclusion threshold intact. The asymmetry is the whole story. Any business with a modest customer list can start suppressing existing buyers from acquisition spend tomorrow. Running positive targeting against your own list, though, demands a database large enough that most small operators won’t clear the bar.

Custom Audiences can be included or excluded at the campaign level, and bid multipliers apply at the ad group level, matching the mechanics advertisers already know from Google Customer Match and Meta Custom Audiences. Conversation context still shapes which ads surface, so list targeting layers onto contextual signals rather than replacing them.

What’s missing is performance data. ChatGPT Ads is young next to Google and Meta’s decade-plus of conversion history, and benchmarks for Custom Audiences haven’t surfaced in any serious volume. OpenAI is building the rest of the stack in parallel, with a growing set of commerce integrations including HubSpot and Shopify, which suggests the platform wants to close the loop from targeting to transaction before advertisers have to ask.

For now, the inclusion floor is the real editorial line. It separates the businesses that can test ChatGPT as a reach channel from the ones that can only use it defensively.

Sources