Attribution windows in ChatGPT Ads: why your conversion count depends on how you count

“How many conversions did the campaign drive?” sounds like a question with exactly one answer. It isn't: the number depends on how long after an ad interaction a conversion still gets credited to the ad — and on which timeline it's counted. ChatGPT Ads now exposes these dials through the API, and in Chatvertise you can flip them directly.

What an attribution window is

The click window defines how many days after a click on your ad a conversion is still attributed to that ad. If someone buys 10 days after clicking, a 30-day window counts it as your campaign's conversion — a 7-day window no longer does. The view-through window works the same way, but for users who saw the ad without clicking.

The options in ChatGPT Ads

• Click window: 7, 14 or 30 days (default: 30).
• View-through window: off or 1 day (default: 1 day).
• Time basis: “ad interaction” buckets the conversion on the day of the click/impression, “conversion” on the day of the conversion itself.

The time basis is subtle but matters for time series: if someone clicks on September 28 and buys on October 3, the conversion lands in your September report on the ad-interaction basis, and in October on the conversion basis. For reconciling with your shop backend (which counts by order date) the conversion basis is often more intuitive; for “what did my September budget deliver?” the ad-interaction basis is.

Why the numbers differ visibly

A real example from one account: with a 30-day click window and a 1-day view window, the same period reports 121 conversions — with a 7-day window and view-through off it's 116, and the two view-through conversions disappear entirely. Neither number is “wrong”; they answer different questions. The tight window shows what your ad triggers directly and quickly, the wide one also captures longer purchase decisions.

How to set it in Chatvertise

On the Conversions page you'll find the new “Attribution & events” section: three toggles for time basis, click and view windows, plus the last 30 days' key figures and a table that breaks conversions down by event — including the click-through vs. view-through split and attributed revenue per event. The selection lives in the URL, so a link to a colleague shows exactly the same view.

Which window should you use?

Short purchase cycles (impulse buys, low-priced products): 7 days shows more honestly what the ad causes. Considered or expensive purchases: 30 days, because decisions take time. And the most important rule: always compare campaigns and periods with the same window — otherwise you're comparing counting methods, not campaigns.