OpenAI this month distributed a product update email to advertisers participating in the ChatGPT Ads Manager Beta, announcing three operational changes to the self-serve platform and a separate early-stage test of dynamic call-to-action buttons within the ChatGPT ad experience. The update, sent by the OpenAI Product Team, was reported on May 22, 2026.

The notice is brief and functional. It covers daily budgets, geo-targeting, and list view totals inside the Ads Manager Beta, then separately outlines what OpenAI describes as a lightweight test of a new ad experience for a small subset of running ads. Together, the changes deepen the operational controls available to advertisers who entered the platform through the self-serve beta that OpenAI opened to all U.S. businesses on May 5, 2026.

What changed in Ads Manager Beta

Daily budgets

The first addition is budget pacing control. According to OpenAI, advertisers creating a new campaign can now choose between a daily budget and a lifetime budget. The distinction matters operationally. A daily budget caps spend per calendar day, making it easier to manage cash flow and prevent a campaign from burning through available funds in the first hours of a cycle. A lifetime budget distributes spend across the full campaign duration, giving the platform more flexibility to optimise delivery timing.

There is a constraint worth noting. According to the update, daily budgets apply only to new campaigns at this stage. Campaigns already running inside the beta cannot be switched to a daily budget retroactively. That limitation suggests OpenAI is rolling out the feature in a controlled manner rather than applying it universally.

For the advertising community, this fills a gap that made early-stage campaign management uncomfortable. Without daily pacing controls, advertisers relying on lifetime budgets had less granular visibility into day-by-day spend behaviour - a friction point in any platform where performance benchmarks are still being established.

Geo-targeting

The second addition is geographic targeting, and it comes with meaningful specificity. According to the update, advertisers can target by U.S. state, designated market area (DMA), or ZIP code. Targeting can be set during campaign creation or updated later in campaign settings.

DMA targeting is a familiar concept in U.S. media buying. The Nielsen-defined geographic zones divide the country into approximately 210 distinct markets, each corresponding to a television viewing area. Advertisers in retail, automotive, and local services have long used DMAs as a standard unit for media planning. Their presence inside ChatGPT's ad interface places the platform's targeting vocabulary in line with what buyers already use across broadcast, streaming, and programmatic channels.

ZIP code targeting offers a finer grain. A single DMA can encompass dozens of ZIP codes, and campaigns targeting hyper-local audiences - a single retail location, a regional event, a geographically concentrated customer segment - benefit from the added precision. State-level targeting, meanwhile, is the coarsest of the three options, suited to national advertisers with state-specific messaging or regulatory constraints.

All three are U.S.-only, which is consistent with the overall state of the ChatGPT ad pilot. The platform launched formally on February 9, 2026, for logged-in adults in the United States on the Free and Go subscription tiers, and geographic targeting at this stage reflects that scope.

The introduction of geo-targeting is a significant step in the platform's maturation. Conversational AI advertising started with contextual matching - ads served based on the content of a user's current conversation, past chat history, and prior engagement - rather than the demographic and geographic layers common to established digital platforms. Adding geographic parameters gives buyers a parallel targeting dimension they can layer on top of conversational context.

List view totals

The third change is less dramatic but practically useful. According to OpenAI, table views inside the Ads Manager now display aggregate totals for key metrics: impressions, clicks, and spend. These totals appear across campaigns, ad groups, and ads in a single view.

For buyers managing multiple campaigns simultaneously, this removes the need to manually sum individual rows to understand portfolio-level performance. Aggregate visibility is a basic expectation in any mature ad management interface - Google Ads and Meta Ads Manager both provide it at every level of account hierarchy. Its addition here is a signal that OpenAI is iterating toward functional parity with established platforms on the reporting side, even as the underlying inventory and targeting mechanics remain distinct.

The ChatGPT ad experience test: dynamic CTAs

Separate from the Ads Manager changes, the update describes a new test affecting the ad experience itself. According to OpenAI, the company is beginning an early test of a new ad experience on a small subset of ads.

Under this test, select ads may include dynamic call-to-action buttons. The four examples provided are "Shop Now," "Book Now," "Sign Up," and "Learn More." These are standard CTA formats familiar from search, display, and social advertising.

What makes this notable is the mechanism. According to OpenAI, CTAs are automatically selected based on the ad creative and the destination experience. The advertiser does not choose the CTA label directly - at least not at this stage. OpenAI states that advertiser controls may be explored in the future, implying that the current implementation keeps that decision inside OpenAI's own system.

The reasoning OpenAI gives is direct. According to the company, this is a lightweight feature designed to help users better understand and engage with ads in ChatGPT. The framing positions the CTA as a legibility aid rather than a conversion mechanism - a way to signal to users what a given ad is asking them to do, in an environment where ads are embedded inside conversational responses rather than displayed in standard banner or search formats.

This design choice has implications. Ads in ChatGPT appear at the bottom of responses, clearly labelled, in a context where the user is engaged with a conversational thread rather than a results page. The CTA adds a directional signal that may help users parse the commercial content faster. Whether that improves click-through rates in a meaningful way will depend on how the feature is rolled out and how widely the subset of ads tested with it is eventually expanded.

Context: where the platform stands

These updates land roughly three weeks after the most significant expansion in the ChatGPT advertising pilot's short history. OpenAI opened its self-serve Ads Manager to all U.S. businesses on May 5, 2026, simultaneously introducing cost-per-click bidding alongside the existing CPM model and publishing a Conversions API for post-click measurement.

That May 5 opening dropped the minimum spend requirement entirely. The threshold had started at $200,000 to $250,000 when the pilot launched on February 9, fell to $50,000 in April, and was removed altogether when the self-serve beta went live. The CPC model introduced at that point carried a recommended starting bid of $3 to $5 per click.

Nine days after the May 5 launch, OpenAI added custom audiences to the Ads Manager, allowing advertisers to upload CSV or TXT files of up to 512MB containing email addresses or phone numbers to create targetable audience segments. That feature, which appeared inside the campaign creation interface at the "Create Ad Group and Ads" step, was a foundational step toward CRM-integrated advertising inside the platform.

The current round of updates - daily budgets, geo-targeting, list view totals, and dynamic CTAs - continues that operational cadence. OpenAI is layering controls in sequence rather than releasing a fully formed product.

The advertising pilot itself launched on February 9, 2026, with a closed roster of large brands and significant minimum spend requirements. CPMs started at around $60 at launch and dropped to as low as $25 by mid-April as more inventory and more advertisers entered the system.

Criteo became the first advertising technology partner in the pilot on March 2, 2026, connecting roughly 17,000 advertiser clients to ChatGPT's Free and Go tiers. By May 5, the technology partner roster had expanded to include Adobe, Kargo, Pacvue, and StackAdapt alongside Criteo - a roster that collectively gives advertisers access to programmatic, retail media, and display buying infrastructure within the platform.

Criteo separately reported in early May that more than 1,000 brands were running active campaigns through its API connection, with AI-referred conversion rates approaching twice those of traditional search in several retail categories.

The geographic targeting addition in today's update is worth reading alongside OpenAI's March 26, 2026 rollout of location sharing inside ChatGPT, which allowed users to share device GPS data for local recommendations and search. That feature was primarily a product change, but it created the infrastructure context in which location-aware advertising becomes more coherent. GPS-derived user location data and geographically targeted ad campaigns now exist within the same platform.

Why these updates matter to advertisers

Each of the three Ads Manager changes addresses a concrete usability gap. Daily budgets solve a pacing problem. Geo-targeting solves a targeting granularity problem. List view totals solve a reporting aggregation problem. None is exotic. All three are standard features in any platform that has been operating for more than a few months at scale.

Their arrival now - about three and a half months after the pilot launched - reflects the operational reality of building advertising infrastructure alongside a product that was not initially designed for it. OpenAI launched ChatGPT in November 2022 without advertising. Plans to test ads were confirmed on January 16, 2026, and the formal pilot began just over three weeks later. The feature velocity since then has been high.

The dynamic CTA test is a different kind of addition. It is not a campaign management control; it is a change to how ads appear to users. Automatic CTA selection based on ad creative and destination is an AI-driven editorial decision about ad presentation. OpenAI flagging the possibility of future advertiser controls suggests the current approach is a test of the mechanism rather than the final product configuration.

For buyers watching the platform develop, the pattern emerging from these weekly updates is one of incremental operationalisation - each release adding a layer that makes the platform more legible to professionals accustomed to managing campaigns across Google, Meta, and programmatic channels. The gaps between ChatGPT's ad infrastructure and those established platforms remain substantial. But the direction of travel is clear, and the pace of feature addition has not slowed since the self-serve beta opened on May 5.

Timeline

Summary

Who: OpenAI, specifically its advertising product team, communicating to advertisers enrolled in the ChatGPT Ads Manager Beta.

What: Three updates to the Ads Manager Beta - daily budget pacing for new campaigns, geographic targeting at the U.S. state, DMA, and ZIP code level, and aggregate totals in list views for impressions, clicks, and spend - alongside a separate early test of dynamic call-to-action buttons (Shop Now, Book Now, Sign Up, Learn More) on a small subset of ads, with CTAs selected automatically based on ad creative and destination.

When: The product update email was distributed on May 22, 2026, approximately 17 days after OpenAI opened the Ads Manager to all U.S. businesses on May 5, 2026.

Where: Inside the ChatGPT Ads Manager Beta, accessible at ads.openai.com, covering campaigns running on the ChatGPT platform in the United States across the Free and Go subscription tiers.

Why: OpenAI is iterating the platform's operational controls toward parity with established ad management interfaces, addressing specific usability gaps in budget pacing, geographic targeting, and reporting aggregation. The dynamic CTA test reflects a separate line of product development focused on improving how users perceive and engage with ads inside a conversational environment.

Share this article
The link has been copied!