OpenAI Tests Ads Manager, Revealing Significant Gaps

OpenAI has rolled out and is testing an ads manager for ChatGPT, a move that converts direct-sold ad inventory into a potential self-serve marketplace. The current interface resembles legacy search ad dashboards but the feature set is preliminary. Advertisers are charged on a pay-per-impression basis for now; cost-per-click and cost-per-acquisition models are marked as coming soon. Targeting and reporting are minimal, limiting performance advertiser adoption while OpenAI iterates.

The product in its present form supports only a small set of advertiser controls and basic reporting. Key capabilities and limitations observed in the test builds include:

• •Charge model: pay-per-impression only; CPC and CPA listed as forthcoming.
• •Targeting: keyword or free-text hints and country-level restrictions; no demographic, interest, or lookalike audience buying.
• •Reporting: impressions and clicks graphs only; no audience size estimates, conversion tracking, or automated optimization.

Platform engineering: The ads manager is evolving rapidly behind feature flags and A/B tests. Recent additions seen in the test include bulk upload support, onboarding screens for new advertisers, and an A/B testing backbone that serves different feature sets to different advertisers. These signals indicate a product-first build with iterative telemetry collection, but key components for performance advertising are not yet present: conversion pixels, attribution windows, audience segmentation APIs, or bidding strategies exposed to buyers.

Privacy and product guardrails: OpenAI has published explicit ad principles, emphasizing mission alignment, answer independence, user control, and privacy. In its public statement OpenAI said, "Ads do not influence the answers ChatGPT gives you." The company also plans to keep Pro, Business, and Enterprise tiers ad-free while testing ads in free and low-cost ChatGPT Go tiers in the U.S.

Building a self-serve ads manager early mirrors Google's Adwords strategy rather than the more typical sequence of ad-first then manager later used by Meta, Twitter, and Snapchat. If OpenAI succeeds, a scalable self-serve channel could materially accelerate revenue and lower reliance on large direct deals. However, most digital ad spend is driven by measurable performance outcomes. Without CPC/CPA buying, audience sizing, conversion tracking, and optimization tools, the offering will be unattractive to demand-side platforms, programmatic buyers, and performance marketers.

The next critical milestones are activation of conversion-based buying models, integration of conversion tracking and attribution, audience sizing APIs, richer targeting (demographic and behavioral), and publisher-side controls for ad quality. Also watch minimum spend thresholds and the availability of programmatic access or API endpoints that let DSPs and ad platforms plug in. Rapid iteration is underway, but the product must close multiple technical and policy gaps before becoming a full self-serve ad ecosystem.