ChatGPT Ads reporting exposes goal and non-goal events

What changed

The relevant documentation supports POST /v1/conversions/insights with include attributed_events. In the reported structure, top-level conversions are goal-only, while nested attributed_event_count can expose non-goal events. Thomas Eccel highlighted this reporting behavior on 6 October; that observation date is not an official feature-launch date and does not establish that a particular interface lacks the same capability.

The distinction matters because a campaign goal is a decision rule, whereas a broader event view is a diagnostic lens. Reporting settings do not change the campaign goal or optimization. Teams should resist the temptation to find a favorable downstream event and retrospectively call it the campaign’s optimized outcome. Instead, define the primary goal, secondary diagnostic events and excluded events before reviewing results. This preserves a clear answer to both questions: did the campaign pursue its intended action, and what other attributed behavior was observed?

Event timing needs careful treatment. Ad-event time is recommended, while non-goal coverage using conversion time is limited. Purchase values can also be null or unavailable. Build reporting that handles those conditions visibly: label the selected time basis, avoid summing missing values as if they were zero, and separate event counts from revenue-based calculations. Where a value field is incomplete, do not present a return-on-ad-spend or profit-style ratio as definitive.

A useful analysis workflow begins with an event taxonomy. Identify the campaign goal, meaningful secondary milestones, duplicate-prone events, refunds or reversals, and events that should not be treated as commercial value. Compare goal and non-goal patterns by cohort or time period only after confirming consistent attribution settings and event availability. If a non-goal event appears unexpectedly high, investigate tagging, funnel changes and time-window effects before changing strategy. The additional detail can improve diagnosis, but it does not by itself prove causality or perfect purchase economics.

Why it matters for advertisers

Broader attributed-event visibility can help advertisers understand the path around a chosen goal, especially when the most immediate event is not the only useful signal. But it also increases the risk of metric shopping. A declared hierarchy prevents secondary events from becoming unexamined success claims and makes reports easier for commercial stakeholders to interpret. The strongest use is as a prompt for validation and experiment design, not as a shortcut to certainty.