ChatGPT Ads is moving from maximum-bid controls to average-cost targets. OpenAI says Target CPA and Target CPC arrive for new campaigns on November 9, 2026, with a separate API transition on November 18 and legacy retirement on January 1, 2027. Advertisers should review campaign meaning and integration behavior before copying old bid values into the new settings.
Checked November 5, 2026. This is a preparation guide for an announced transition, not a claim that every new control is already available. GPTAd is independent of OpenAI.
Three dates to put in the handoff
- November 9, 2026: Target CPA replaces Max Bid for new conversion campaigns; Target CPC becomes available for new click campaigns. Existing campaigns can begin moving to the new options in Ads Manager.
- November 18, 2026: When legacy API fields are used to create a campaign or change an existing ad group's bid type, Max Bid selects Target CPA and Max CPC selects Target CPC; the submitted amount becomes an average-cost target. Until then, those fields keep their existing behavior.
- January 1, 2027: Existing Max Bid and Max CPC campaign controls retire, and the legacy API fields are deprecated. OpenAI says those fields remain supported through December 31, 2026.
The current official bidding guide also renames Maximize Results (Autobid) to Intelligent bidding. Dedicated target API fields are being added; partners can adopt them when available. The November 18 behavior described above is scoped to the specified operations, not every possible edit.
Replace the meaning, not just the label
A target is an average-cost goal. It is neither an individual auction ceiling nor a guarantee that the achieved average will stay below it. Individual outcomes and the eventual average can cost more or less. The bidding goal also remains separate from the event that triggers billing.
For a team reviewing a migration, this changes the approval conversation. A number previously approved as a maximum bid should not silently become an approved average cost-per-outcome goal. Record what the amount means, the selected outcome, the billing event and the spending budget together. Our budget-planning guide explains how to separate business affordability from platform settings.
A practical preparation checklist
The following is GPTAd's recommended review process, not an additional OpenAI requirement.
- Inventory affected configurations. Separate new-campaign templates, existing campaign settings and scripts that change an ad group's bid type. Give each a named owner and a planned review date. A saved template can contain an outdated assumption even when no campaign is currently running.
- Keep a before-and-after record. Capture the objective, selected conversion event, current strategy, amount, currency, billing event and budget. Add a plain-language explanation of the proposed target so finance and the operator approve the same economics.
- Check the evidence behind the target. Compare like-for-like reporting periods and the same conversion event. Distinguish observed campaign performance from the business's desired acquisition cost. A desired margin is a planning input, not evidence that an auction will deliver it.
- Verify the measurement path. Confirm that the campaign is connected to the intended source and that real events reach it. Review the selected event, duplicate handling and reporting delays before treating a cost change as a bidding problem. Our conversion-event checklist helps structure that review. Event quality and event volume answer different questions; a high quality score alone does not establish enough conversions to scale.
- Review the result after a supported change. Read back the saved strategy and amount, then evaluate spend, outcomes and actual average cost together. Decide in advance who can pause a test if tracking breaks or the economics become unacceptable. Restrictive targets, limited eligible opportunities or weak conversion signals can leave budget unspent. Do not interpret that as an instruction to increase it automatically.
API teams: treat November 18 as a semantic change
At this review, the developer bidding guide and Ad Groups reference still describe legacy controls. The Help Center gives the transition dates but does not provide exact names for the new target fields. We therefore do not publish speculative request examples or assume that a UI label is an API parameter.
For your integration review, identify code, validation rules and user-facing labels that assume a maximum. Add tests for the intended economic meaning as well as the accepted value and response. Before implementation, recheck the current reference for exact field names, units and supported operations. A successful request alone does not establish that the system interpreted a value as the ceiling your team intended.
The useful deliverable now is a short migration record: affected configuration, owner, intended target, measurement check and verification plan. Keep it beside the spending approval so the platform change does not silently change the team's assumptions.
