Export the uplift report

Exports the uplift report, a campaign’s or a journey’s holdout metrics, to a CSV file, through the same task lifecycle as message statistics export: call export, poll status, call result, then download.

Not the control group export

TASK_TYPE_EXPORT_CONTROL_GROUP exports a control group’s member list: devices and user counts. TASK_TYPE_EXPORT_CONTROL_GROUP_ANALYTICS, documented here, exports the uplift report’s metrics instead: no member list, no device data. They’re separate task types with separate arguments.

Request

POST https://api.pushwoosh.com/api/v2/statistics/exportMessagesStatistics/export

Use "type": "TASK_TYPE_EXPORT_CONTROL_GROUP_ANALYTICS" and an export_control_group_analytics object instead of export_messages_v2. Authentication and the rest of the lifecycle (status, result, lastTasks, delete, download) work exactly as described on message statistics export.

export_control_group_analytics accepts:

Name	Required	Type	Description
application_code	Yes	String	Application code.
date_from	Yes	String	Window start over the report’s own timestamp, YYYY-MM-DD or YYYY-MM-DD HH:MM:SS. A date-only value expands to the start of that day.
date_to	Yes	String	Window end, same format. A date-only value expands to the end of that day.
campaign	No	String	Campaign code, exports that campaign’s holdout only. Mutually exclusive with journey.
journey	No	String	Journey UUID, exports that journey’s Holdout Split points only. Mutually exclusive with campaign.

Leave both campaign and journey empty to export every campaign and every journey the application has a report for.

Important

The window can’t exceed 90 days. A wider window isn’t trimmed: the request fails with 400 and the message exceeded the maximum date interval. Max interval: 90 days. The uplift report itself keeps 90 days of rows, so a wider request couldn’t return anything more anyway.

Example request

{
  "type": "TASK_TYPE_EXPORT_CONTROL_GROUP_ANALYTICS",
  "export_control_group_analytics": {
    "application_code": "XXXXX-XXXXX",
    "date_from": "2026-08-01",
    "date_to": "2026-08-31"
  }
}

result returns export_control_group_analytics_result.file and .file_url, the same shape as message statistics export’s result.

What’s in the file

One row per campaign or per journey holdout point, per goal event. The underlying report writes a fresh row every day for the same pair, and the export keeps only the newest one, so you get the current numbers, not a history of every past run.

Column	Description
timestamp	When this row was computed.
owner_type	TYPE_CAMPAIGN or TYPE_JOURNEY.
owner_id	The campaign or message code for a campaign row, the journey’s UUID for a journey row.
point_uuid	The Holdout Split point’s UUID for a journey row. Empty for a campaign row.
event_name	The goal event this row measures.
attribution_window_seconds	The attribution window applied, in seconds.
treatment_users, control_users, treatment_conversions, control_conversions, treatment_events, control_events	Raw counts behind the metrics below.
uplift_pct, incremental_events, percent_of_treatment, z_score, p_value, confidence_pct	Same metrics as the uplift report.
significance	TYPE_SIGNIFICANT, TYPE_NOT_SIGNIFICANT, or TYPE_NOT_ENOUGH_DATA.
revenue_tracked	1 if the goal event’s revenue source was configured, 0 if the row is only guessing from a legacy fallback. See Revenue in the report.
treatment_revenue_<currency>, control_revenue_<currency>, incremental_revenue_<currency>	Three columns per currency, lowercase currency code (for example treatment_revenue_usd). The file only gets columns for currencies that actually appear in the exported rows.

참고

When revenue_tracked is 0, every revenue cell in that row is empty, not 0.00. A blank cell means the source wasn’t configured, not that the send made no money.