Attribution
Lookback window
Also called attribution window.
Lookback window is how far back from a conversion the model searches for touches to credit. Touches older than the window are ignored.
How it is measured
The window is a number of days, set per model or per touch type. The system sets a start date for each conversion by subtracting the window length and considers touches from that date onward.
Match it to the buying cycle. Compute the median days from first touch to purchase and set the window to cover most buyers, not all of them.
Worked example
A cabinetry maker finds that the median gap from first visit to deposit is 41 days and the 80th percentile is 73. Their analytics default is 30 days.
At 30 days, first-click credit for the home-design magazine ad is 22 deposits. With a 75-day window it is 49. Nothing about the ad changed. The tool simply began to see the earlier touch.
How it differs
A lookback window defines how far back touches are searched. A conversion window defines how long after a touch a conversion still counts. They are two sides of the same limit.
Common errors
Keeping the default. Using the same window for cheap and expensive purchases. Changing it between reports. Setting a window longer than the data you have. Comparing tools with different windows.
In practice
Compute your own time-to-purchase distribution and choose the window from it. Record the number in the report footer. Revisit if the product or price changes.
See also
Conversion window, First-click attribution, Multi-touch attribution