OpenAI today took what it described as "the next step" in its ChatGPT advertising pilot, rolling out a beta self-serve Ads Manager that allows businesses in the United States to sign up and purchase ad placements directly, without working through a holding company or agency partner. The announcement, published May 5, 2026, also introduced cost-per-click (CPC) bidding alongside the existing cost-per-mille impressions model, and confirmed the recent launch of a Conversions API and pixel-based measurement tools. Together, the changes represent the most significant operational expansion of the ChatGPT ad platform since the pilot formally began on February 9, 2026.

The self-serve interface is available at ads.openai.com and is being opened gradually. Businesses interested in access can sign up for an account and wait for OpenAI's team to complete verification before gaining full platform access. Campaigns will not deliver until all required setup steps are complete, according to OpenAI's help documentation.

What the Ads Manager beta offers

According to the Ads Manager Beta Overview published by OpenAI, the platform supports three core functions: creating and managing campaigns either individually through a guided workflow or at scale through bulk upload; monitoring performance through table views, charts, and CSV exports; and managing account settings including members, permissions, API keys, billing details, and change logs.

The campaign structure follows a three-tier hierarchy. A Campaign defines the objective, budget, start and end dates, and targeting parameters. An Ad Group is a collection of ads organized around a specific theme or intent area, and it is at this level where advertisers supply context hints - phrases that describe the types of conversations or topics where their products may be relevant. An Ad is the individual creative unit shown in ChatGPT, consisting of a title, description copy, image asset, and landing page URL.

Ad format is standardized. According to OpenAI's help documentation, each ad unit includes the advertiser name, a favicon (brand logo), a title headline, description copy, a landing page, and an image asset. The documentation recommends a headline of approximately 16 characters and description copy of approximately 32 characters. Crucially, it specifies that logos should not be used as the primary visual element in the image creative - a design constraint that pushes advertisers toward product or lifestyle photography.

Account setup happens in five steps: accessing Ads Manager Beta and creating an account, completing account onboarding and verification, completing account information (including uploading a favicon), setting up billing and payment, and inviting team members. Each business requires only one account owner to create the advertiser account, and that owner can subsequently invite additional team members. If an agency or operator manages multiple advertisers, each advertiser requires its own separate account.

Billing must be fully configured before campaigns can deliver. This requires creating a billing profile with a business name, address, and invoice delivery email, and adding a credit card with associated billing address information. OpenAI's system reviews new advertiser accounts and emails the account holder once full platform access is available.

CPC bidding and its implications

The introduction of cost-per-click bidding marks a shift in how OpenAI positions its advertising inventory. In the initial phase of the pilot, which launched February 9, CPM-only pricing was the sole available option, set at $60 per thousand impressions - a rate that drew comparison to premium streaming and NFL broadcast inventory and stood well above Meta's typical CPMs, which often fall under $20.

According to OpenAI's announcement, the shift to CPC as an additional option was motivated by the nature of ChatGPT user activity. "Many ChatGPT conversations are active and decision-oriented," the announcement states. "People are often learning about a category, comparing options, or deciding what to do next. In those moments, a click can be a meaningful signal that an ad was relevant and helped someone move forward." Advertisers are only charged on a click outcome under the CPC model.

The mechanics differ by objective. According to the "Ads in ChatGPT: The Basics" documentation, advertisers choose a Reach objective to buy on a CPM basis or a Clicks objective to buy on a CPC basis. For CPC campaigns, advertisers can set custom maximum bids, and OpenAI recommends a starting maximum bid of $3 to $5 USD per click. The default maximum bid for CPM campaigns remains $60. The underlying auction uses a relevance-weighted, second-price mechanism designed to maximize value for both the advertiser and the user.

OpenAI has confirmed it will continue supporting both CPM and CPC buying simultaneously. The platform intends to add further bidding models over time, aligned with the outcomes advertisers want to optimize for.

Measurement: Conversions API and pixel

One of the more consequential additions in this announcement is the launch of Conversions API and pixel-based measurement tools. The measurement gap had been one of the most persistent criticisms levelled at the ChatGPT ad platform since the pilot's inception. As PPC Land reported in April 2026, the platform still lacked a pixel or standard attribution infrastructure at that stage, leaving advertisers unable to measure conversions with the precision they expected from established channels.

According to OpenAI's announcement, the Conversions API and pixel tools allow advertisers to measure what happens after someone engages with a ChatGPT ad - including purchases, leads, sign-ups, and other defined actions. Performance data is provided in aggregated form. Individual conversation data is not shared with advertisers.

Ads Manager Beta reporting now includes impressions, clicks, spend, click-through rate (CTR), average CPC, average CPM, and conversions (the last contingent on having conversion measurement configured). These metrics can be reviewed in three formats: the campaign/ad group/ad table view, trend charts over time, and CSV reports that can be downloaded for daily or cumulative values across a selected date range.

Advertisers can also track ChatGPT ad traffic through static tracking parameters such as UTM parameters appended to landing page URLs. These parameters persist on ad clicks and work with existing analytics tools.

Ad selection and delivery mechanics

OpenAI's ad selection system does not operate on exact-match keywords. According to the platform documentation, ads are selected primarily based on relevance to the context and intent of the conversation. The system considers multiple inputs: context hints provided by the advertiser at the ad group level, the landing page content, the ad title, and the ad copy. Context hints help guide matching but do not guarantee delivery in specific conversation types.

This is a fundamental structural distinction from search advertising. In Google Ads, an advertiser bids on specific keywords and triggers. In ChatGPT, the advertiser describes - rather than specifies - the conversational territory where their products may be relevant. The platform's inference engine makes the final matching decision.

Ads appear below relevant ChatGPT conversations, not within the responses themselves. OpenAI has consistently maintained that advertising does not influence the answers ChatGPT provides. According to the documentation, "Ads are clearly identified in the experience" and "Ads remain distinct from ChatGPT's responses."

Regarding audience eligibility, according to OpenAI's documentation, ads are currently shown to Free and Go users in the United States, Canada, Australia, and New Zealand. The international expansion to Canada, Australia, and New Zealand was announced on March 26, 2026, the same day OpenAI disclosed the pilot had crossed $100 million in annualized revenue within six weeks. Users on Plus, Pro, or any Business plan are excluded, as are users who indicate or appear to be under 18 years old.

Partners and the broader ecosystem

Today's announcement also clarifies the role of OpenAI's growing partner network in the advertising stack. OpenAI states it has been collaborating with major agency holding companies including Dentsu, Omnicom, Publicis, and WPP to support businesses purchasing ChatGPT ads. These holding companies were among the first confirmed partners when the advertising program was formally confirmed on January 16, 2026.

On the technology partner side, OpenAI lists Adobe, Criteo, Kargo, Pacvue, and StackAdapt as current participants. Criteo was formally identified as the first ad tech partner on March 2, 2026, connecting approximately 17,000 advertiser clients to ChatGPT's Free and Go tiers. Through these technology partners, advertisers can access ChatGPT ad inventory using tools and processes already integrated into their existing workflows. According to OpenAI, partners help support campaign budgeting, bidding, and ad creative, while OpenAI's own system retains full control over all delivery decisions.

The self-serve Ads Manager is positioned as the route for companies of all sizes - from small businesses to global brands - that want to operate independently rather than through an agency or technology partner. The pilot has already attracted more than 600 advertisers, with minimum spend thresholds cut from $250,000 at launch to $50,000 in April 2026, and today's announcement makes the pathway even more accessible.

Brand safety and user privacy

OpenAI's policy positions ChatGPT ad placements as brand-safe by design. According to the platform documentation, ads are placed only near conversations that are safe, appropriate, and aligned with user trust. The safeguards are designed to prevent ads from appearing in sensitive user contexts, with specific exclusions defined in OpenAI's Ads Policies.

User privacy protections are built into the measurement architecture. The aggregated reporting structure means that an advertiser receives performance data about campaign impact without receiving access to the content of individual user conversations. This is a structural commitment rather than a settings-level option. OpenAI's U.S. privacy policy, updated on April 30, 2026, formalized that it receives purchase data from advertisers and shares user information with outside marketing partners for targeting purposes - disclosures that now sit in binding legal language for the first time.

What this means for marketers

The opening of a self-serve interface addresses a gap that has constrained the ChatGPT ad pilot since February. Until now, buyers without direct OpenAI relationships or agency partnerships had no mechanism to participate. PPC Land has covered the platform's rapid evolution from a tightly controlled pilot with steep entry requirements toward something resembling an open advertising market.

The addition of CPC bidding is particularly significant for performance marketers. CPM models require an advertiser to trust that impressions at a given price will generate outcomes. CPC shifts the risk: the advertiser pays only when a user clicks. For direct-response campaigns - lead generation, e-commerce, subscription sign-ups - CPC is the preferred model because it ties spend directly to the first measurable user action. The recommended starting bid of $3 to $5 per click is competitive with mid-market search CPCs on Google Ads, though comparisons are difficult given the different intent signals at play.

The Conversions API addresses what analysts described as the most structurally important gap in the platform's commercial case. As PPC Land noted in its coverage of the April CPM decline, the absence of independent conversion verification left advertisers "taking OpenAI's word for it" on campaign outcomes. The arrival of a pixel and Conversions API gives advertisers a mechanism to close the attribution loop between a ChatGPT ad click and a downstream action on their own properties.

The three-tier campaign structure - Campaign, Ad Group, Ad - mirrors the architecture familiar from Google Ads and Meta Ads Manager, which lowers the operational learning curve. The context hint system is the principal novelty: rather than selecting keywords, advertisers write natural language descriptions of relevant conversations. That is a different skill than keyword research, but one that performance marketers working with broad match and audience signals may find conceptually adjacent.

Timeline

Summary

Who: OpenAI, the company behind ChatGPT, announced the expanded availability of its self-serve Ads Manager beta, CPC bidding, and conversion measurement tools. The platform is accessible to businesses of all sizes in the United States, from SMBs to global brands. Agency partners including Dentsu, Omnicom, Publicis, and WPP, and technology partners including Adobe, Criteo, Kargo, Pacvue, and StackAdapt, are also part of the ecosystem.

What: OpenAI today launched an open beta of its self-serve Ads Manager for US advertisers at ads.openai.com, adding cost-per-click bidding (recommended starting bid: $3 to $5 per click) alongside the existing $60 CPM model, a Conversions API for post-click measurement, and pixel-based attribution. The platform follows a Campaign/Ad Group/Ad hierarchy with context hint targeting at the ad group level. Reporting now covers impressions, clicks, spend, CTR, average CPC, average CPM, and conversions.

When: The announcement was published May 5, 2026. The broader ChatGPT advertising pilot began February 9, 2026. The self-serve Ads Manager was first made available to a limited subset of advertisers in April 2026, with today's announcement marking its broader rollout.

Where: The Ads Manager beta is available in the United States at ads.openai.com. ChatGPT advertising currently reaches Free and Go tier users in the United States, Canada, Australia, and New Zealand.

Why: OpenAI is building advertising infrastructure to fund its broader AI mission. The company faces an estimated $14 billion in projected 2026 losses and carries an internal advertising revenue target of $2.4 billion for the year. The shift from a closed, direct-sales model to self-serve tooling follows the same operational path taken by Google, Meta, and Amazon as their advertising platforms scaled - and allows OpenAI to widen the advertiser base beyond the large-budget brands and holding company relationships that defined the pilot's first phase.

Share this article
The link has been copied!