“I approved the ad, so I know what’s running.” With ChatGPT Ads creative enhancements enabled, I would not make that bet. OpenAI lists October 30, 2026 as the date enhancements become the default for new and existing self-serve and BPO campaigns, while preserving existing opt-outs and excluding sales-managed accounts. That makes the handover question concrete: who checks what can change, preserves the settings you intended, and verifies the path from ad to conversion? Getting the campaign live is not the same as getting it under control.
Myth 1: “The copy I approved is the only copy that can run”
I get why this one sticks. You write a headline, submit it for approval, and reasonably expect that headline to be the ad. Approval sounds like the end of the editing process. With creative enhancements on, it is not.
OpenAI lists four changes the system can make:
- New or updated headlines and descriptions tied to the conversation.
- Translation into another language.
- Image resizing to fit the format.
- Destination URL optimization.
Those versions draw on the conversation context, your existing ad, and landing page content. They can run automatically without per-variation review or a preview of every version. The copy you approved is an input, not necessarily the final wording someone sees.
The advertiser email quoted by PPC Land makes the control problem particularly clear: one Creative enhancement toggle covers all four changes. There is no per-type control that lets you lock the headline while allowing image resizing, for example. “I’m fine with a smaller image” and “please rewrite my offer” sit behind the same switch. Convenient, in a very platform-shaped way.
So I would not start by rewriting the approved copy a fifth time. I would check the campaign’s enhancement setting first, because campaign settings take priority over the account default. Then I would establish what can actually be inspected, rather than assume every variation is available for review.
Check permission to change the copy before troubleshooting the copy. Otherwise, you are editing a draft someone else is still allowed to revise.
Myth 2: “The destination URL cannot change automatically”
This is the assumption I would expect from careful advertisers: the people who matched the ad to a landing page on purpose. OpenAI reviews each ad with its landing page and expects the two to match. It is an easy jump from “this page was reviewed” to “this page is fixed.”
It is also the wrong jump. Destination URL optimization to a relevant page on your site is one of the four enhancements. The same toggle that permits creative changes permits destination optimization.
The awkward part is what the supplied guidance leaves unexplained. Neither the advertiser email nor the help text explains how the replacement page is chosen, whether it must stay on the same domain or path, or whether advertisers can exclude pages. PPC Land flags that page-selection gap, along with questions about logo cropping.
Search Engine Journal adds two measurement gaps: no reporting that separates enhanced versions from originals, and no explanation of how UTM tags carry over when the destination changes.
That does not prove your tags are being stripped or your traffic is going to the wrong page. It means neither deserves to be treated as settled just because you entered a URL during setup.
I would check the path in this order:
- Confirm the destination URL entered in the ad.
- Verify the page reached through the ad where you can observe it.
- Check whether the tracking tags and click reference survive the trip.
If a shopper reaches a generic category page instead of the tested product page, you have a different journey to diagnose. If tracking disappears, you have a measurement problem to diagnose. Neither gets fixed by polishing a headline.
Check destination behavior before blaming the creative. The page is part of the ad’s job, not an administrative detail attached to it.

Myth 3: “The October 30 change wipes my old opt-out”
The rollout messaging gives this myth plenty to feed on. The advertiser email says Friday, October 30; the help page briefly displays October 29, then returns to October 30. Early email language says self-serve only, while the updated help page includes self-serve and BPO. When the announcement shifts around, assuming the settings shift with it is understandable.
But existing account- and campaign-level opt-outs are preserved, and campaign settings take priority over the account default. A default change is not a blanket instruction to overwrite every previous choice.
What is true instead is less dramatic and more useful: memory is not an account audit. “We opted out” does not tell me whether that choice lives at account level, campaign level, or both. “The account is opted in” does not tell me whether an individual campaign overrides it.
I would verify the hierarchy rather than argue about the announcement:
- Open the account-level default.
- Check each campaign’s enhancement setting.
- Confirm which setting controls the campaign you are investigating.
An old campaign-level opt-out still takes priority over an account-level opt-in. That is exactly why looking at only one screen can produce the wrong diagnosis.
Verify an existing opt-out; do not assume the default change erased it. And if nobody has touched the setting, do not mistake silence for a deliberate choice.
Myth 4: “Every ChatGPT Ads account follows the same default”
“Automatic rewriting is on by default” makes a tidy headline. It also leaves out the part that decides whether the change applies to you.
The October 30, 2026 default applies to new and existing self-serve and BPO campaigns, not sales-managed accounts. Account arrangement comes before setting advice.
This is where a DIY setup can go sideways without anyone doing anything especially foolish. You follow a checklist written for one account arrangement while operating under another. Then a provider diagnoses the checklist instead of the account.
Before paying for help, I would ask three plain questions:
- Is this account self-serve, BPO, or sales-managed?
- What is the account default set to?
- Which campaigns override that default?
The answer should describe your account, not repeat the rollout headline. If a prospective provider cannot answer those questions in a screen share, keep looking.
Account type first, settings second, copy last. It is not glamorous troubleshooting, which is often a good sign.
Myth 5: “Once the campaign is live, setup is finished”
A campaign can accept spend without giving you a usable picture of what that spend achieves. Live status does not certify tracking, destination behavior, or ownership of the next change. I would treat it as a delivery milestone, not a completed handover.
The basic setup constraints still matter. Country, currency, and time zone are permanent; delivery requires live billing; and the daily minimum is $25 USD. Ads show to Free and Go plans. The same setup guidance describes a spend-reporting lag of 7 to 8 hours and a conversion lag of 24 to 48 hours. An empty report is not, by itself, a diagnosis.
That creates two different traps. One advertiser assumes live means everything works. Another sees no immediate data and starts changing things before the reporting catches up. Neither has established what needs fixing.
I would treat setup as four checks that belong together:
- Tracking. Set up Pixel and/or Conversions API before launch, preserve the
opprefclick reference through redirects, and allow OAI-AdsBot. Installing tracking and preserving the reference are parts of the same job. - Landing pages. Confirm that the ad and page match. OpenAI reviews them together; the conversion path needs the same attention.
- Creative settings. Verify the enhancement toggle, the account-to-campaign hierarchy, and destination behavior. “Copy approved” is not a substitute for any of those checks.
- Budget and ownership. Establish the spending guardrails, which changes need approval, and who reviews the account weekly.
The last check is easy to skip because it sounds like administration. It is actually the difference between a setting someone chose and a setting everyone assumes someone else chose.
A campaign with live status and dead tracking is not a finished setup. It is spending. Before calling the launch complete, name the person responsible for checking what happens next.

Myth 6: “Hiring help means starting over from scratch”
This is the myth I find hardest to kill because it lets both sides avoid the uncomfortable work. The DIY advertiser assumes a provider will wipe the account and rebuild, so they delay asking. The lazy provider is happy to oblige: a rebuild bills clean hours and avoids the less flattering question of whether anyone diagnosed the problem.
A useful handover starts with the account you have. Which toggle is on, and where? Which campaign overrides the default? Which destination loads? Where does the click reference disappear? Those are diagnostic questions. “Shall we rebuild it?” is a proposal looking for a diagnosis.
There is also a practical distinction between types of help. A one-off setup hands back a built account. Consulting advises you but leaves implementation with you. Hands-on management operates inside the account week by week. If your problem is keeping track of what changes after launch, another setup alone leaves you holding the same ongoing job.
Ownership matters here. The setup guidance recommends that you create the advertiser account yourself, invite the agency as Member, keep separate accounts per advertiser, and name one approver for copy and claims. OpenAI reviews each ad with its landing page, so the approver needs to understand both. Keeping the account yours makes later change control possible without making the provider its owner.
I would ask for an account-specific diagnosis before authorizing anything new. Not an account-shaped sales deck. The actual settings, destinations, and measurement path.
Send a diagnostic packet, not just “spend is flat”
If you ask, “Can anyone set up my ChatGPT ads for me?”, the quality of the answer depends partly on what you send. “Spend is flat” leaves plenty of room for somebody to sell you new copy. Settings, destinations, and tracking give a hands-on provider something concrete to investigate.
I would put these four things in one email:
- Account arrangement and settings: self-serve, BPO, or sales-managed, with screenshots of the account default and each campaign override for creative enhancements. Campaign settings take priority, so screenshots beat memory.
- Toggle state and timing: whether creative enhancements are enabled and when that setting changes relative to October 30. Remember that the single toggle covers all four enhancement types; it is not a menu of individual permissions.
- Destinations in and out: the URL entered, the page users actually reach where observable, and whether UTMs and the click reference survive.
- Measurement and spend: Pixel and/or Conversions API status, billing status, daily cap, and any observed conversion-reporting delay. Nobody should mistake a 24-to-48-hour lag for proof of failure.
That packet helps distinguish management work from a specific fix or a reporting delay. It also gives you a way to judge the provider. Do they explain the account you sent, or immediately propose replacing it?
Hand over responsibility, not just access
What I want for a stuck DIY account is hands-on ownership of the work inside a client-owned account, with written rules for who approves changes and who stops spend. I would also check access before paying for more help. A recommendation you cannot implement is still your problem.
That is the operating-model distinction I care about with groas: continuous execution inside agreed guardrails, with a named strategist accountable for the outcome, rather than a deck of recommendations for you to implement alone. A one-off setup leaves ongoing checks with you. Advice-only consulting leaves execution with you. Neither solves a handover problem simply by calling it support.
No rebuild theater. No assumption that outsourcing fixes a broken offer. Ask for a diagnosis before you buy a replacement. Live is not settled. Someone has to verify what can still change after launch, and that someone should be the provider you pay.

