ChatGPT Ads Manager: understand the interface before you spend

Ads Manager, the single interface for ChatGPT Ads

OpenAI concentrates all ChatGPT media buying in one product, Ads Manager (Beta), at ads.openai.com. There is no other documented route: no console inside the regular ChatGPT account, no separate advertiser interface for large accounts.

One vocabulary point up front, because it causes a lot of confusion: Ads Manager is not the same thing as the ad account. Ads Manager is the tool; the ad account is the entity that carries the country, currency, billing and users. One user can open Ads Manager and find several ad accounts in it. Creating the account itself is covered on our ChatGPT Ads account page.

One thing to note about that account right away: the creation screen locks in three values you will not be able to change later, the country, the currency and the time zone. Currency determines the applicable minimum daily spend, since OpenAI publishes a table of minimums by currency; time zone determines where a day starts and ends in reporting, and therefore how your curves read and how the daily budget paces; country is the advertiser entity's country, and should match the documents you provide at verification. If any of the three is wrong, the only documented remedy is to create a new account.

The product is still explicitly labelled beta. In practice that means two things: features appear without notice, and some screenshots you will find online no longer match the interface.

The three-level hierarchy: campaign, ad group, ad

Every campaign in Ads Manager follows the same structure: campaign, then ad group, then ad. The documentation describes it in exactly that order, and the hierarchy is not optional: you cannot create an ad without an ad group, or an ad group without a campaign.

This three-level architecture is the same one Google Ads and Meta Ads Manager use, which is why the tool feels immediately familiar. That is also the trap: the same words do not cover exactly the same settings from one platform to the next. The detailed comparison lives on our ChatGPT Ads vs Google Ads and ChatGPT Ads vs Meta Ads pages.

A good campaign, per the documentation, groups together the ad groups and ads that support a shared business goal. That is the tie-breaker to remember whenever you hesitate between creating a new campaign and creating a new ad group.

Which level controls what

The table below is the mental map to keep open through setup. Looking for a setting at the wrong level is the most expensive beginner mistake, because it sends you hunting where the control does not exist.

Two points deserve emphasis, because they cut against habits picked up elsewhere:

• The bid is entered at the ad group level, not the campaign: a maximum CPM bid for a Reach campaign, a maximum CPC bid for a Clicks campaign.
• The budget is steered at the campaign level, either as a campaign total or as a daily budget. Details and minimums are on our ChatGPT Ads pricing page.

Roles and permissions: Global Admin is not Ads Admin

This is the subtlety that blocks the most teams on setup day. Global Admin and Ads Admin are separate roles. A Global Admin manages the tenant, identity settings and users, but being a Global Admin does not automatically grant permission to create an ad account. For that, the Ads Admin role has to be assigned explicitly to the people who need it.

Once the ad account exists, rights are granted at a second level: each user or group receives an ad account role that controls what they can do inside that specific account, such as viewing or managing campaigns.

Another useful distinction: ChatGPT workspace membership and Ads access are managed separately. Ads-only customers manage their users in the Global Admin Console, while access to a given ad account still has to be assigned inside Ads Manager.

OpenAI also documents SSO and SCIM setup for Ads Manager, which matters to organisations that provision access from a directory.

Running several accounts from one email address

The documentation is explicit: a single email address can access multiple ad accounts, as long as that user has been assigned a role in each one. The user then switches between accounts inside Ads Manager.

Do not mistake that for a centralised management account. It is not an equivalent of a Google Ads MCC or a Meta Business Manager: there is no pooled billing, no consolidated performance view, and no copying of assets from one account to another. It is a plain account switcher sitting on top of permissions granted one at a time.

For an organisation running several advertisers, the implications are heavy enough to deserve their own page: see ChatGPT Ads for agencies.

Reporting: what the interface gives you

Ads Manager Beta reporting currently includes impressions, clicks, spend, click-through rate (CTR), average CPC, average CPM and conversions.

Those metrics are available in three forms:

• table view, to compare campaigns, ad groups and ads row by row;
• chart view, to watch a metric move over time;
• CSV export, in daily or cumulative values across the selected date range.

The export pulls the same campaign, ad group and ad rows you see in the table, so you can feed them into a spreadsheet, a BI tool or your own attribution model.

One warning that is worth real money: the main Conversions column counts click-through conversions only. View-through conversions, shown for some accounts on a fixed one-day window, are separate supplemental reporting that OpenAI says should not be added to conversions or used for core metrics like CPA. The measurement detail sits on our pixel and conversion tracking page.

Reading delivery statuses, and the dreaded "Not serving"

Ads must pass review before they can serve. Once campaigns are submitted and approved, the status shown in Ads Manager tells you whether delivery is actually happening.

When an ad shows Not serving, the documentation tells you to hover over the status to see the reason. The most common documented causes:

• the ad account has not been verified, and a banner in Ads Manager flags it;
• the campaign budget limit has been reached;
• the campaign end date has passed;
• there is an issue with the payment method.

The good news: many blockers clear by editing and resubmitting the ad or campaign. Some ads, though, are rejected on policy grounds, or stop serving while account review, business verification or a billing action is outstanding. Rejection grounds are covered on our ChatGPT Ads policies page.

The targeting available from the interface

Targeting is set at the campaign level and controls where ads can deliver. Ads Manager Beta supports:

• country targeting, at campaign level;
• in the United States only, finer targeting by state, designated market area (DMA) and ZIP code.

On top of that sit custom audiences, built from lists of customers or prospects (email addresses or phone numbers). Once created, an audience can be used for campaign-level inclusion, limiting eligibility to people in that audience, or campaign-level exclusion, preventing delivery to them.

The rest of your targeting does not run through checkboxes but through the context hints entered at the ad group level, which describe the conversations where your offering is relevant. They are broad thematic signals, not exact-match keywords, and they deserve their own method: see ChatGPT Ads context hints.

Product feed campaigns

Ads Manager handles a distinct campaign type for commerce: campaigns built from a product feed. The principle is to connect a catalogue, mark which products are eligible for advertising, and let the platform build ads from catalogue attributes.

For a retailer this is the fastest route to volume, since it removes the need to write one ad per product. It brings its own constraints on feed format and freshness, detailed on our ChatGPT Ads for ecommerce page.

The interface-level takeaway here: the feed is connected from inside Ads Manager, and the public advertiser API does not expose endpoints to create a feed connection or upload a catalogue file.

What Ads Manager does not do yet

Worth stating plainly, since it saves hours of hunting through menus. As of this update, OpenAI's documentation does not describe:

• a multi-advertiser management account along the lines of MCC or Business Manager;
• pooled billing across several ad accounts;
• demographic targeting by age or gender;
• lookalike audiences built automatically from an existing audience;
• retargeting of past site visitors;
• a product feed API to automate catalogue connection.

We word these as gaps in the documentation, not as permanent prohibitions: the product is in beta and any of them could appear. Just be wary of content claiming they already exist.

Getting started: the order of operations

A sober sequence that follows the order the tool imposes:

1. Create the ad account and check country, currency and time zone twice.
2. Finish verification of identity and business, then set up billing: without it, nothing delivers.
3. Assign roles, separating who administers from who runs campaigns.
4. Put measurement in place before the first campaign, not after: with no measured conversion, oCPC is out of reach.
5. Create the campaign: objective, budget, dates, targeting.
6. Create the ad group: theme, context hints, bid.
7. Create the ads, then submit them for review.
8. Wait at least 24 hours after launch before concluding you have a delivery problem.

Before step 7, run your landing page through the ChatGPT Ads landing page checker: blocking OpenAI's user agents is enough to sink an otherwise flawless ad.

What is not publicly documented

Points we flag as unconfirmed for lack of an official source:

• The full list of ad account roles and the exact rights attached to each are not reproduced here.
• The maximum number of ad accounts one email address can access is not quantified.
• Review turnaround times for an ad or a campaign are not the subject of any public commitment.
• The reporting history depth and any export limits are not specified.
• No guarantee of delivery, volume or outcome is offered by OpenAI, and nobody can offer one on its behalf.