OpenAI Ads connector
AnsweredWhich business application?:
OpenAI Ads, via the OpenAI Events API — OpenAI's conversion / measurement
API for their newly launched ads product. Angi is a launch partner and
is building a revenue feed to send total-value conversion signals to
OpenAI so their bidder can optimize toward value rather than raw
conversion volume. Anticipated channel spend is $125K+/month at launch,
growing from there.
What object in the destination?:
A conversion event. Each event carries:
- Conversion timestamp (UTC, ISO-8601)
- Event type
- Pixel / account identifier
- Currency (ISO-4217, per event)
- Value (non-negative decimal; zero-value conversions included)
- OpenAI-issued click identifier captured on landing
- Consent / opt-out flag (opted-out users suppressed at source)
Grain: one row per conversion, where each conversion corresponds to a
Service Request tied to a paid OpenAI ad click.
How will records be identified?:
Primary identifier is the OpenAI-issued click ID captured on landing —
unique per click-to-conversion. Deduplication should be at click-ID
grain: one conversion event per click ID, with the latest value winning
on updates. If a composite key is preferred, click ID + our internal
Service Request ID is also unique.
What sync behaviour?:
- Cadence: hourly ideal; 24-hour freshness is the minimum (parity with
our existing Google Ads revenue feed).
- Semantics: append new conversions as they occur, with support for
value updates on previously-sent conversions (total value is an
estimate that can be revised — OpenAI supports restatements).
- Hard partner constraint: OpenAI rejects conversions with a timestamp
more than 7 days old. The sync must drop or flag stale rows rather
than retry indefinitely against that ceiling.
- Volume: ~1–5M conversion events per month at maturity for our account.
- Auth: API key / Bearer token per advertiser account.
Additional details:
-
Official comment
Hi Eric,
This is a valuable request and the OpenAI Ads connector idea has been added to our feature improvements backlog. I am curious to better understand your specific use case for this connector. What reporting or optimization challenges are you aiming to solve through a Fivetran integration with the OpenAI Events API? Any additional context about your expected workflows or target destinations would help as we evaluate feasibility and design options.
Are you looking for reporting and insights data mainly?
Updates and progress will be shared with the community on this thread as they become available.
Thanks,
Tom
Please sign in to leave a comment.
Comments
1 comment