What it is
The Attributed conversions dataset runs your conversions through the same attribution engine the report builder uses, and unloads the un-aggregated result: each row is a conversion joined to a touchpoint that earned credit toward it.- Grain: the attribution model decides. Single-touch models (first/last/platform/non-direct) emit exactly the winning touchpoint — one row per conversion. Multi-touch models (linear, u-shaped) emit one row per touchpoint with its fractional credit, so a conversion’s rows sum to its full value.
- Windowed on: the conversion’s
event_time_utc(UTC timestamp). - Path segment:
attributed_conversions.
Two independent windows apply. The export window bounds which conversions are included. The
separate attribution window (a setting below) bounds how far back a touchpoint may sit
before its conversion to still earn credit.
Settings
Configured once at creation (immutable, like the rest of the export):enum
required
first_touch, last_touch, first_platform_touch, last_platform_touch, last_non_direct_touch,
linear, linear_paid, or u_shaped. Determines the row grain (see above).enum
required
How far back a touchpoint may sit before its conversion:
1, 7, 14, 28, 30, 60, or 90
days, or LTV (no limit).enum
required
Attribute by profile (the contact) or by company.
array
Optional source filter. Empty means all sources (the report-builder default).
Columns
Attributed conversion fields
Conversion metadata
The visit-context record for the conversion. Each column is prefixedconversion_. All optional.
Touchpoint metadata
The visit-context record for the attributed touchpoint (the ad-click visit). Each column is prefixedtouchpoint_. All optional.
For a multi-touch conversion, the conversion metadata repeats across that conversion’s
touchpoint rows, while the touchpoint metadata differs per row.