Adobe Campaign
This page covers delivery data from Adobe Campaign v8. To connect Subsets to Campaign so it can write experiment assignments, see the Adobe Campaign integration guide.
Subsets measures who was treated from your actual send logs, not from workflow targeting counts, so typology and pressure rules that drop recipients after targeting are accounted for.
What to deliver
Share either the raw Campaign data below or a curated equivalent (a sends, opens and clicks table plus a delivery table carrying the delivery code). Subsets does all joins itself.
| Data | Campaign source | Notes |
|---|---|---|
| Sends | nms:broadLogRcp | One row per targeted recipient, including those never sent. Include @status and the failure reason so Subsets counts only sent messages (Adobe) |
| Opens and clicks | nms:trackingLogRcp | Opens are url/@type 2, clicks are url/@type 1 (Adobe) |
| Delivery catalog | nms:delivery | With deliveryCode, label and internalName, for every delivery including those a recurring delivery creates on each run |
| Person identifier | The log rows, or a mapping table | The same customer ID your profile feed sends Subsets and that Subsets writes to customerId. If the logs carry only Campaign's recipient ID, add a recipient ID to customer ID mapping table |
Leave out email addresses and other personal data; the customer ID is all Subsets uses.
The delivery code is the join key
Subsets links each send to an experiment treatment by its delivery code, matched exactly against the codes registered in the experiment. Export deliveryCode unmodified: same spelling, same casing. Subsets ignores surrounding spaces and nothing else.
Delivery logs that reach Subsets only through the Adobe Experience Platform source connector are not supported today: that path does not carry the delivery code.
Timestamps and cadence
- Timezone. Deliver event timestamps in UTC, or tell Subsets the timezone and type. Attribution compares send times with the time Subsets wrote the assignment, so an unstated local-time offset shifts sends near that boundary.
- Cadence. Deliver at least daily, partitioned on the event timestamp, and re-send a trailing window of recent days so late opens and clicks arrive. A stalled export looks exactly like an experiment with no sends, so tell Subsets who monitors it.
- Retention. By default Campaign keeps delivery logs for 6 months and tracking logs for 1 year, and these can vary per implementation (Adobe). Land the logs continuously; history past the purge cannot be backfilled.
- Transport. Campaign extracts files to Amazon S3, Azure Blob storage or SFTP (Adobe). The step from there into the warehouse you share with Subsets needs a named owner on your side.
Data quality
- Enable open and click tracking on experiment deliveries, with no excluded links. Untracked links read as non-clicks.
- Keep proofs, seeds and Adobe control-group rows out of the export, or flag them; they inflate treated counts.
- Continuous deliveries (one delivery reused indefinitely) work for attribution but collapse the per-email breakdown into one bucket.