What Are ChatGPT Ads Audience Lists, and Should Your Brand Test Them Now?

ChatGPT Ads just added the one feature that turns it from a media experiment into a real performance channel: first-party audience targeting. Here is what you need to know and what to do about it.

Last week, something slipped into ChatGPT Ads with almost no fanfare. Under the “Tools” section in the Ads Manager, advertisers started seeing a new option called “Audiences.” No big press release from OpenAI. Just a feature that lets you upload your customer list, using raw or hashed emails and phone numbers, and use it as a filter on your campaigns.

This matters. Not because audience targeting is new, but because ChatGPT Ads has been strictly contextual until now. Every ad on the platform was matched to what the user is asking about, not who the user is. Adding first-party data changes that logic entirely. You are now able to influence not just what context triggers your ad, but which people see it.

Why First-Party Data Changes Everything

The reason agencies spend so much time managing customer match lists in Google and Meta is that first-party data converts better than cold audiences. Always. When someone who already bought from you, or who fits your exact ICP, sees your ad in a high-intent context, the economics look completely different from a generic campaign targeting broad interests.

ChatGPT Ads was already showing promise before this update. Criteo, running campaigns through the platform via its API connection, reported AI-referred conversion rates approaching twice those of traditional search. That figure applies specifically to retail verticals like consumer electronics and home goods, and it comes from a single vendor, so treat it as directional rather than gospel. But the platform’s blended click-through rates of around 3.8 to 4 percent across early accounts, with a CPC range of roughly $1 to $4, are competitive with Google and Meta from the start.

Now layer first-party data on top of that. You are no longer just showing up when someone asks a question your product is relevant to. You are showing up when a specific person, someone you already have a relationship with or know is a strong prospect, asks that question. That combination of context and identity is what makes customer match targeting so powerful on Google, and it is now available on the fastest-growing ad platform of 2026.

What OpenAI Actually Rolled Out

On July 7, 2026, advertiser Craig Graham spotted and shared a screenshot on LinkedIn showing the new Audiences feature inside ChatGPT Ads Manager. Barry Schwartz at Search Engine Roundtable and Search Engine Land both covered it that day. The feature appears under “Tools” in Ads Manager and prompts users to “Create your first audience: Upload raw or hashed emails and phone numbers to use as audience filters for campaigns.”

That is the scope for now. No lookalike audiences, no behavioral interest segments, no CRM integrations. Just a list upload and a filter. But that is actually the right place to start, and it follows the pattern of how platforms in this category tend to mature.

The broader context matters here. ChatGPT Ads launched in February 2026. OpenAI dropped the minimum spend requirement and opened its self-serve Ads Manager to all US businesses in May. CPC bidding came in April. CPA bidding, the conversion-optimized campaign type, rolled out June 5. Now, in July, first-party audience lists are live. The platform has gone from brand-only CPM buys at $60 CPM to a full performance channel with first-party targeting in five months. According to Search Engine Land, it is already generating $100 million in annualized revenue for OpenAI. That is not a test anymore.

How to Set Up Audiences

Getting your list ready

OpenAI accepts raw or hashed email addresses and phone numbers. Best practice is to hash before uploading using SHA-256, which is the same standard Google and Meta use. Start with your cleanest, most relevant segment rather than your full database. If you have 50,000 contacts, begin with the 5,000 who have converted in the last 90 days. You want a tight signal, not volume.

Setting realistic match expectations

Every platform matches only a subset of uploaded lists, because they can only connect contacts who have accounts with matching email or phone data on file. ChatGPT Ads has not published match rate benchmarks publicly yet. Treat this like you treated Customer Match when Google first launched it: assume a conservative match rate, set your test budget accordingly, and refine from there as data comes in.

Using it as a campaign filter

Based on how the feature is described in the platform interface, audience lists function as a filter layered on top of contextual targeting, not a replacement for it. Think of it the same way you would use Customer Match in Google Ads as an observation or bid modifier layer rather than a standalone campaign strategy. The contextual match still drives when ads appear. You are narrowing the eligible audience, not redefining the trigger.

Strategy Angles Worth Testing

Exclusions for new customer acquisition

Upload your existing customers and exclude them from acquisition campaigns. This is often the most immediate return on any platform because you stop wasting budget retargeting people who already bought from you. Set this up before anything else.

Retention and upsell targeting

Build a separate campaign layer targeting known customers with a specific offer or cross-sell message. ChatGPT’s conversational context is well-suited for this because users are often in an active problem-solving mode. If someone is asking ChatGPT about a challenge your product already solves for them, being present in that moment is a high-value placement.

Suppression lists as a foundation for lookalikes

There are no lookalike audiences on ChatGPT Ads yet. But if OpenAI follows the same product roadmap as Google and Meta, they are coming. Getting a clean, well-segmented seed list uploaded now means you will be ready to activate lookalikes the day they drop, rather than scrambling to build one under pressure.

Common Mistakes to Avoid

Uploading your full unfiltered database

Bigger is not better here. A matched audience of 500 highly relevant customers will outperform a matched audience of 10,000 loosely related contacts in most cases. Segment before you upload. Your best customers, recent converters, and high-LTV accounts should be separate lists so you can learn what each group does differently on this platform.

Treating this exactly like Google Customer Match

ChatGPT Ads is still contextual at its core. The audience list filters who sees your ads. It does not change when or why the ad shows. Your creative still needs to fit the conversational context of ChatGPT. Ads that read like display banners will not work here regardless of how precise the targeting is. Write copy that fits how people actually talk when asking ChatGPT for advice or a recommendation.

Skipping conversion tracking setup

If you have not set up the ChatGPT Ads pixel or Conversions API yet, audience list targeting will not help you optimize toward outcomes. You will be improving reach without improving measurement. Sort tracking first, build your audience lists second, then layer CPA campaign optimization on top.

What This Means Going Forward

ChatGPT Ads has hit an inflection point. In five months OpenAI went from basic CPM ads to a self-serve performance channel with CPC, CPA, and now first-party audience targeting. The platform is moving faster than Google and Meta did in their early years, partly because it can learn from their playbook and partly because the user base is already enormous and highly intent-driven.

The agencies building their ChatGPT Ads infrastructure now, including conversion tracking, audience lists, and a working understanding of what creative tone actually lands in a conversational interface, will have a compounding head start. That is how it worked when Google Shopping launched. That is how it worked when Meta added custom audiences in 2013. The early adopters who put in the foundational work when the platform was underpriced captured a window that closed quickly.

At Incisive Growth, we are already running ChatGPT Ads as part of multi-channel paid media strategies for clients in competitive categories. If you are running Google Ads and you have a clean first-party data list, ChatGPT Ads with audience targeting is worth a structured test budget this quarter. The cost of not testing is higher than the cost of a modest, well-measured experiment.

Frequently Asked Questions