Attribution Windows ("Lookback Window")

What is an Attribution Window (Lookback Window)

An Attribution Window is the period of time between a marketing event (ex clicks, view) and a conversion event (ex Purchase) in which a specific marketing channel is eligible to receive credit for driving the conversion.

For example, if I apply a 220 day attribution window:

  • A user who clicks on a Google ad today and converts 200 days from now will have a Google touchpoint on their path to conversion.
  • Similarly, a user who clicks on a Google ad today and converts 300 days from now will NOT have a Google touchpoint on their path to conversion, since the clicks falls outside the 220 day window.

What Attribution Windows Does Rockerbox Apply

Rockerbox attempts to apply the maximum attribution window to all channels, to allow tracking of the maximum number of marketing touchpoints on a user's path to conversion.

This is a different approach than what exists in the social platforms or DSPs, which enable you to limit the lookback window (and apply different lookback windows to view vs click based activity).

The window applied by Rockerbox depends on multiple factors, including how long deterministic data is available and the time period of the data that can be provided by our partners.

A breakout of the attribution windows applied per marketing channel is below:

Type of Touchpoint
Example Channels
Attribution Window


Paid Search

Paid Social

Organic Search

Organic Social





220 days

Deterministic Views




90 days

Linear TV

All Linear TV

14 days



7 days

Direct Mail

All Direct Mail

220 days

Direct Partnerships




7 day click / 1 day view*

*Attribution Windows may be more limited for Direct Partnerships due to limitations on the data each partner is able to provide.

Marketing Event to Session Limits

For OTT and Linear, we also impose a limit on eligible marketing events based on the time between the ad and web session.

  • OTT: defaults to 300 minutes
  • TV: 5 minutes

We enforce these limitations to prevent over-attribution due to the probabilistic nature of both attribution methods.

How did we do?