ChatGPT Ads Conversion Warnings: What to Check Before Raising Spend

ChatGPT Ads now puts conversion event quality recommendations beside conversion activity in Ads Manager. For an ecommerce team already running campaigns, that changes the next budget conversation. Check whether purchase events are arriving cleanly and whether campaigns are optimizing for conversions before treating a weekly performance change as a reason to spend more.

An OpenAI advertiser update shared by Search Engine Roundtable describes a redesigned Conversions page with event volume and trends, the share of active campaigns optimizing for conversions, and warnings that link to Diagnostics. The trade report was published October 19. The email image itself does not show a send date. It also says active advertisers receive weekly performance emails. These views make a measurement check easier to start. They do not establish that a reported conversion was caused by an ad.

What should you decide before increasing spend?

Decide whether the events used to judge the campaign are complete, timely and mapped to the purchase outcome you care about. Then decide whether enough eligible products are serving to make the test meaningful. A week over week change is a prompt to investigate, not a verdict when the underlying events or product set changed.

This is a different decision from installing a pixel before a first test. The new Conversions overview gives a team with an active campaign a place to inspect ongoing event health. Use it to separate a real change in campaign performance from a change in what Ads Manager can observe. Keep a dated baseline for both your event counts and the products actually serving.

Start with the warning, then trace the purchase

Open Conversions and note the event volume trend, any event quality recommendation, and the active warnings linked to Diagnostics. Record the exact warning, its affected event and when it appeared. Compare that period with your store's order records and analytics. The numbers need not match because attribution rules and observed sessions differ, but a sudden divergence merits investigation before a budget decision.

Check the purchase event path end to end with your media buyer or analytics owner. Confirm that the event represents a completed order rather than a checkout start, that the same order is not counted twice when browser and server signals both arrive, and that the order value and currency are sensible. Test consent and checkout paths relevant to your store. If an event is missing, find the break in the instrumentation rather than rewriting product copy to explain a reporting gap.

Next, look at the share of campaigns optimizing for conversions. The new overview exposes that share, according to OpenAI's email, but it does not tell you which optimization setting is right for your account. Inspect each active campaign's stated goal and the event it uses. If one campaign is optimized for a different action, keep its results separate from the purchase campaign when you review cost per acquisition.

Compare the same product set and the same measurement window

A campaign may have sound conversion tracking while its product set changes beneath it. Export or record the products selected for the test, feed status, availability and price at the start of the comparison window. Check whether a blocked offer, stock change or stale description reduced the set that could serve. A shift in product mix can move both clicks and purchases without proving a change in ad effectiveness.

Ask your media buyer to compare like periods and hold the attribution window constant. Look at click through conversions and any view through reporting separately. A view through number is not a second confirmed order count, and adding it to click through conversions may double count people across views. The new screen is an operational aid, not a substitute for a controlled incrementality test.

Use the new weekly email as a cue to repeat this review, not as the review itself. If Ads Manager reports a week over week improvement, check whether event volume, warnings, optimization settings and eligible products remained comparable. If those inputs changed, annotate the result. The practical question is whether the team can explain the movement well enough to make the next allocation decision.

What does the visual ad update change?

OpenAI's advertiser email also describes experiments with wider cards, larger desktop images and fuller width mobile images. It says these tests use existing campaigns and creative and require no action from advertisers. A separate multi-image format is planned for an initial US test with select advertisers during image generation. Do not treat those experiments as a new placement available to every merchant or as evidence that a different image will convert better.

Keep your product images accurate and useful for the offer being advertised. Review the existing creative against the landing page and feed, especially variant, availability and current price. That is ordinary campaign hygiene. Wait for account-specific eligibility and results before making a production-wide creative change in response to a format test.

What can wait until the diagnostic is clear?

OpenAI says it expanded measurement partner options and added a way to report bugs from Ads Manager. A new partner name alone is not a reason to rebuild your reporting stack. First identify the decision your present setup cannot answer, whether it is missing purchase events, disputed channel credit or a need for a holdout. Then evaluate the partner's actual account support and setup requirements with your analytics owner.

Keep paid results separate from organic AI visibility. An ad impression and a cited product in an unpaid answer are different outcomes. Track spend, clicks and attributed purchases for ads. For shopper questions, inspect whether your products appear in answers and which pages are cited. A product information fix may serve both workstreams, but a paid campaign result cannot stand in for evidence of an organic recommendation.

The useful next step is a short decision log: the warning and affected event, the comparable campaign and product set, the observed purchase evidence, and the action owner. Paz experts choose campaign products from your priorities, clear product feed blockers, and write ad titles and descriptions from your sources. We ship within approved guidelines, bring exceptions to you, and review per-product performance with your media buyer against a baseline. Talk to us about the product information work behind a test you can judge.