The default is not a bid

You did not choose the bid strategy. The builder did.

For eligible new ad groups in ChatGPT Ads, Maximize results arrives preselected. OpenAI’s own Help Center page is blunt about what that means: the system automatically sets and adjusts bids to get as many results as possible from your budget, optimizing toward the campaign goal you already picked (clicks or conversions). It prioritizes result volume. It does not guarantee delivery against a specific CPA, CPC, ROAS, or other cost-efficiency target “at this time.” If you need a bid limit, you switch to Manual: Max bid in the ad group’s Bid strategy section (OpenAI Help Center, Maximize Results Bid Strategy, updated around August 2026).

That is not a quiet UI polish. That is the platform deciding that spend-the-budget-for-count is the normal path, and that a hard ceiling is the exception you must remember to click.

PPC Land covered the same documentation and the August product email on September 26, 2026. The load-bearing sentence is the default itself. There is no account-level preference that forces Manual once and holds. The choice is per ad group, every time. Operators who skim the Bid strategy step inherit volume bidding and then wonder why efficiency looks like a random walk.

What actually changed

Three facts matter. Do not blur them into “ChatGPT got Smart Bidding.”

Fact one: Maximize results is selected by default for eligible new ad groups. OpenAI’s setup path is explicit. Create a campaign in Ads Manager Beta, continue to the ad group step, review Maximize results as the preselected option, choose Manual: Max bid if you want a different strategy, then finish the ad group and launch. The same default shows up in product-feed campaign setup: in the Bid strategy section, Maximize results is selected by default for eligible new ad groups, and ChatGPT Ads automatically adjusts bids to help get as many clicks or conversions as possible within budget based on the campaign goal (OpenAI Help Center, Create Campaigns from Product Feeds, updated September 26, 2026). Standard campaigns and catalog campaigns share the same OS.

Fact two: volume is the promise; efficiency is not. The Help Center limitation is the consequential paragraph, not a footnote. Maximize results prioritizes result volume and does not guarantee CPA, CPC, ROAS, or other cost-efficiency targets at this time. PPC Land’s September 26 explainer correctly separates that from Target CPA, Target ROAS, and Cost Cap style strategies elsewhere in the market. Those express a number the system is meant to hold. Maximize results expresses count. The budget is the constraint. Whatever cost per outcome falls out of the auction is the cost per outcome. The phrase “at this time” is a roadmap signal without a date, and you should treat it as unfinished product, not as a soft promise you can bank in a QBR.

Fact three: Manual Max bid is the only documented hard ceiling, and the API exposes the same split. On the Ads API, Maximize Results is the name for maximize_clicks and maximize_conversions. Both use billing_event_type: "click", both require a daily campaign budget, both omit max_bid_micros, and both require an Idempotency-Key on create. Fixed bid is the path that needs max_bid_micros. You can switch strategies by updating bidding_config, but Maximize Results cannot ride a lifetime budget. Switch the campaign to daily first, then change the ad group (OpenAI Developers, Bidding & Budgets, accessed September 26, 2026). UI Manual: Max bid and API fixed_bid with an explicit max are the same control idea: you name a ceiling. Maximize results refuses that field on purpose.

The mechanism: budget is the constraint

Once Maximize results owns the bid, the daily budget is doing almost all of the governance work. That would already be a sharp change. ChatGPT Ads also moved fixed daily budgets to seven-day averages in July 2026 without an advertiser opt-out (PPC Land, September 26, 2026).

OpenAI’s Daily Budgets Help page (captured September 26, 2026; page stamp about twenty-two days prior) defines the math. A daily budget is the average amount you want to spend per day over a seven-day period. Spend on a given day can run higher or lower. Unused amounts may be used later in the same seven-day window. Maximum daily spend is twice the daily budget. Maximum seven-day spend is seven times the daily budget. If you change the budget mid-day, that day’s max is based on the highest daily budget active at any point that day. Mid-week changes prorate the seven-day limit from the change through the next Sunday at midnight.

Pair those rules with a strategy that has no CPA or CPC target and no max_bid_micros. You get an operating system that is allowed to chase count, flex daily spend up to 2x, and still call the plan “on budget” as long as the seven-day envelope holds. Two of the three classic levers (bid and daily hard stop) sit with the platform. The third (the average budget number) is what you still type.

Measurement makes the trap tighter. PPC Land notes that view-through conversion reporting landed in the same August window with a fixed one-day window, and OpenAI walled that metric off from the main conversions total, from CPA, from billing, and from bidding. Post-view credit can rise in a column while Maximize results optimizes toward conversions the bidding system can actually see. Reading view-through as proof that automation is “working” is reading a number the bid strategy cannot use.

More self-serve and reseller volume is walking into that same default. MARKETECH APAC reported on September 25, 2026 that Criteo expanded its ChatGPT Ads integration across six SEA markets (Malaysia, Indonesia, the Philippines, Singapore, Thailand, Vietnam), taking the reseller path to sixty-one markets globally. That is color, not the thesis. The thesis is what those new ad groups inherit when nobody clicks Bid strategy.

What operators get wrong

Mistake one: skipping Bid strategy because the rest of the form looked finished. The default is selected for eligible new ad groups. There is no account preference that locks Manual forever. Building the fourth ad group at 5:40 p.m. without opening Bid strategy is how you accidentally run volume bidding on a brand-safety or margin-sensitive SKU set. Treat the Bid strategy pane like a required field even when the UI treats it as already answered.

Mistake two: reading Maximize results like Target CPA with a friendly name. It is not. OpenAI says it prioritizes volume and does not guarantee cost-efficiency targets at this time. If your internal brief says “automated bidding toward a $40 CPA,” and the live ad group is Maximize results with no Manual max, the brief and the account disagree. Rename the strategy in your docs as volume bidding until a real efficiency target ships.

Mistake three: judging day-one CPC or CPA as if the daily budget were a hard daily stop. Under seven-day averaging, a day can spend up to 2x the daily budget. A spike is not automatically “broken pacing.” It can be legal pacing against a weekly envelope. Conversely, a quiet day is not proof the strategy is thrifty. Evaluate Maximize results on a rolling seven-day cost-per-result and on whether the conversion signal underneath is clean, not on a single Tuesday screenshot.

Mistake four: maximizing conversions on a thin or changing pixel. Conversion-objective Maximize results is only as sound as the event the campaign optimizes for. ChatGPT Ads measurement has been moving month to month (pixel, Conversions API, advanced matching defaults). If the event is noisy, duplicated, or mostly post-view theater, volume automation will happily buy more of the wrong thing. Fix the event definition before you celebrate a cheaper CPA that is really a dirtier denominator.

Mistake five: confusing Manual Max bid with “I still control efficiency.” A max bid is a ceiling on what you will pay per billed event under fixed bidding. It is not a CPA target, not a ROAS floor, and not a promise the auction will clear near your number. It is still the only documented hard ceiling in this stack. Use it when you need a stop. Do not pretend it is Google-style tCPA. And do not leave Maximize results running on a lifetime-budget campaign in the API path: the docs require daily budget for Maximize Results, and switching strategies without fixing budget type is how create/update calls fail while the UI still looks calm.

Distinct from the last month of Vault tickets

This essay is not a rematch of recent Vault media and commerce pieces. Say the fence once.

Not bill-is-not-the-click: that ticket was oCPM billing and impression versus click economics. Today’s ticket is bid strategy default and volume automation, not which event you pay for. Not country-is-not-the-playbook (SEA and Taiwan geo open). Not catalog-is-the-campaign (Shopify catalog craft). Not upgrade-is-not-a-toggle or search-term-is-not-the-journey (Google AI Max). Not click-is-the-conversation, crm-is-the-ads-console, pipe-is-not-the-placement, or checkout-is-not-the-storefront.

Today’s ticket is simpler and meaner: when you create an eligible new ad group, the platform has already voted for spend-the-budget-for-count unless you override it.

If your working group is still arguing impression billing under a “Maximize results” subject line, you are in the wrong thread.

What to do this week

Run this as a bid-strategy and budget-governance sprint, not as a feature celebration.

• Open Bid strategy on every live ChatGPT Ads ad group created since mid-August. Record Maximize results versus Manual: Max bid in a simple sheet. Anything that launched on the default without a written decision gets an owner and a deadline.

• Decide strategy by margin and signal quality, not by fashion. Use Maximize results only where volume inside a known budget envelope is the actual goal and the conversion (or click) signal is trustworthy. Use Manual: Max bid wherever you need a hard ceiling, a test cell with controlled CPC, or a brand-sensitive set that cannot absorb auction spikes.

• Rewrite the brief language. Ban internal phrases that imply Maximize results holds CPA or ROAS. Replace them with “volume bidding within daily budget” until OpenAI ships a documented efficiency target. Align agency SOWs to the same vocabulary.

• Rebuild pacing alerts for seven-day averages. Alert on trailing seven-day spend versus 7x daily budget, and on trailing seven-day cost per result, not only on single-day overspend versus the daily number. Remember the legal 2x daily max before you page someone at midnight.

• Separate view-through from bid truth. Keep the one-day view-through column out of the Maximize results scoreboard. Optimize and judge on the conversions (or clicks) the bidding system is allowed to use.

• API and feed builders: set strategy explicitly. Do not omit bidding_config.strategy and hope. For Maximize Results, use daily budget, omit max_bid_micros, send Idempotency-Key, and match maximize_clicks or maximize_conversions to the campaign objective. For a ceiling, use fixed_bid with max_bid_micros. Confirm the returned bidding_config after every create or switch.

• Product-feed campaigns get the same audit. The feed setup docs preselect Maximize results too. Catalog scale without a bid decision is still a bid decision. Filter sets and custom labels do not replace Bid strategy.

• Write the override ritual. Because there is no account-level “always Manual” switch, put Bid strategy into your QA checklist next to geo, budget, and conversion event. New joiners and Friday afternoon launches are where defaults win.

Close

Defaults are policy with better UX.

ChatGPT Ads did not remove Manual: Max bid. It made volume automation the path of least resistance for eligible new ad groups, on a channel whose cost benchmarks are still thin, with daily budgets that already flex across seven days. Operators who treat Bid strategy as optional will inherit a count machine and then grade it like an efficiency product.

The default is not a bid. It is a decision you still owe.