Classify
Name the current relationship, recency, product, geography, permission, and suppression status.
Source basis: OpenAI custom-audience and ad-group guidance · July 29, 2026
ChatGPT Ads customer lists: a lifecycle playbook
A list becomes useful only after the business separates who should be reached, who should be suppressed, who may deserve a different bid, and who should not be uploaded at all.
25,000matched users are currently required before use
100,000+matched users are recommended by OpenAI
0.1x to 10xdocumented ad-group bid multiplier range
New versionis required when the audience membership changes
00
The direct answer
Separate the business condition before you make the audience file.
A mixed customer export hides the decision that the campaign needs.
OpenAI documents custom audiences built from customer or prospect email addresses or phone numbers. Once an audience is ready, it can be used for campaign inclusion, campaign exclusion, or an ad-group bid adjustment. Those controls have different jobs. They should not be applied to the same undifferentiated list simply because the identifiers are available.
Start with the customer condition and the desired business action. A recent buyer may need to be excluded from acquisition. A lapsed subscriber may be eligible for a specific return offer. A qualified prospect may support a narrow campaign only when the offer and permission basis fit. A suppression record should remain suppressed.
List size answers “Can it deliver?” Lifecycle answers “Should it?”
Decide
Choose inclusion, exclusion, bid adjustment, or no upload for that condition.
Prove
Compare campaign delivery with new-customer, repeat-customer, revenue, or qualified-lead records.
01
Lifecycle groups
Use a customer condition that the business can recognize later.
The group name should be recoverable from the CRM, store, or subscriber record.
Qualified prospects
People who meet a defined condition but have not completed the target purchase, booking, or contract.
- Source and permission known
- Offer still relevant
- No active suppression
Current customers
People who already completed the action the acquisition campaign is meant to produce.
- Recent purchase or contract
- Product ownership known
- Cross-sell separated
Past customers
People whose last qualifying action falls beyond a written recency window and who remain eligible for contact.
- Lapse definition written
- Return offer appropriate
- Recent service issues checked
| Condition | Likely decision | Evidence after launch |
|---|---|---|
| Recent customer in a new-customer campaign | Exclude unless the campaign is built for a distinct additional product. | New versus existing customer rate, duplicate acquisition cost, product ownership. |
| Qualified prospect with a current open opportunity | Decide with the sales owner whether paid inclusion helps or disrupts the active conversation. | Opportunity stage, sales contact, ad exposure, final outcome. |
| Lapsed customer with a defined return offer | Use a separate campaign or ad group with a message that acknowledges the next relevant need. | Reactivation, order value, margin, unsubscribe, and complaint rate. |
| Suppressed, disputed, refunded, or ineligible record | Do not use for acquisition or reactivation. Preserve the suppression rule. | Suppression coverage and zero unintended activation. |
02
Three controls
Give each audience one job in the test.
Inclusion, exclusion, and a bid adjustment are not interchangeable labels.
Use the smallest control that answers the question.
Include: limit a campaign to a defined audience when the test is specifically about that known group.
Exclude: prevent a defined group from entering a campaign where its presence would waste spend, distort acquisition, or contradict a suppression rule.
Adjust the bid: change the ad-group maximum bid for members of a ready audience. OpenAI documents multipliers from 0.1x to 10x. Write the reason before choosing the number.
Do not upload: keep a record out when permission, ownership, accuracy, lifecycle status, policy fit, or business purpose is unresolved.
The useful sentence
Write this before upload: “For people in this condition, we will use this control because we expect this business outcome, and we will stop if this evidence appears.”
If the sentence cannot be completed, the audience is not ready.
03
Version control
Make replacement part of the operating rhythm.
OpenAI says custom audiences cannot be edited after creation.
Use a name that records the lifecycle condition and source date without exposing personal information. Upload one identifier type per audience version. Wait for processing and inspect the final matched size. A large source file can still produce a matched audience below the current 25,000-user minimum.
When membership changes, create a replacement. Confirm that it reaches ready status, attach it to the intended campaign or ad group, then archive the old audience. Keep the source rule, file owner, permission owner, campaign use, and replacement date together.
04
Lifecycle check
Complete the business decisions before the upload.
These checks separate a usable audience from a large file.
Customer-list readiness
0 of 12 ready05
Primary sources
What this playbook is based on.
These official pages were checked on July 29, 2026.
Identifier types, thresholds, inclusion, exclusion, bid multipliers, processing, replacement, and archiving.
Context hints, audience controls, separation of products and use cases, and maximum bids.
Current content, targeting, product, and advertiser boundaries.
Does the 100,000-user recommendation mean the audience is good?
No. It is a platform size recommendation, not proof of relevance, permission, customer quality, or profitable delivery. The lifecycle condition and final business outcome still need to be evaluated.
Are context hints the same as a customer list?
No. Context hints describe the conversations where an ad may be useful. A custom audience is built from advertiser-provided identifiers. They can work together, but they answer different targeting questions.
Use the customer record with a clear purpose.
Stan Consulting connects audience planning, paid ads, CRM records, landing pages, and customer-quality reporting.