Attribution Windows Compared: 1 Day, 7 Day, and 28 Day
The window is the quiet setting that decides which channel wins the report. What each common window actually measures, and how to choose one without fooling yourself.
Every attribution argument is secretly an argument about windows. Shorten the window and the last click channels look efficient. Lengthen it and the prospecting channels that started the journey weeks ago finally get their credit. Neither report is lying. They are answering different questions while wearing the same title.
This guide compares the three windows most stores actually use, states what each one is honest about, and gives a way to pick based on the buying cycle rather than on which result the team prefers.
The 1 day window: closest to the till
A one day click window credits what happened right before purchase. It is the most conservative common setting and the closest to provable influence, which makes it useful for tight budget control and for spotting channels that harvest demand created elsewhere.
Its blindness is structural: consideration that takes longer than a day becomes invisible, so considered purchases, higher price points, and anything researched across a weekend are systematically undervalued.
The 7 day window: the working compromise
Seven days captures a realistic research cycle for most consumer purchases: see it, think, compare, return, buy. It is the default for a reason, balancing provable proximity against the way people actually shop for anything above an impulse price.
It still undervalues slow categories. Furniture, appliances, and considered fashion regularly run two to four week journeys, and a seven day window hands their originating channels credit to whichever retargeting ad happened to be last in line.
The 28 day window: the fullest and the loosest
A twenty eight day window sees the whole journey for most categories, including the first touch that genuinely started it. It is the fairest view of prospecting and content, and the most generous, because a month of exposure contains coincidences the window happily counts as influence.
Use it for strategic questions, channel mix, prospecting value, content investment, and distrust it for weekly budget moves, where its generosity turns into noise and double counting across platforms becomes severe.
Choose by buying cycle, then hold it steady
Measure the real time between first visit and purchase in your own data, and pick the shortest window that covers most of that journey. A window matched to behaviour is a measurement choice. A window chosen because a channel looks good under it is a marketing choice wearing a lab coat.
Whatever you choose, hold it steady and label every report with it. Comparisons across time are only meaningful under a constant window, and the most common attribution error in growing stores is not the window itself but quietly changing it between the reports being compared.
Related reading
If this topic is useful, these earlier guides go deeper on connected questions: Reading Attribution During Short Flash Sales Without Fooling Yourself, Attribution When Shoppers Switch Devices Before Buying, and Planning Incrementality Tests You Can Actually Run After Peak.
Frequently asked questions
What does an attribution window actually decide?
How far back before a purchase the system looks for touches to credit. The window decides which channels appear effective, which makes it one of the most consequential settings in the whole measurement stack.
Which window is right for most stores?
The shortest one that covers the real buying cycle in your data, often seven days for everyday consumer goods and longer for considered purchases. The right answer comes from journey length, not from preference.
Why not always use the longest window?
Because generosity becomes error: a month of touches includes coincidences, inflates prospecting credit, and multiplies double counting between platforms. Long windows suit strategic questions, not weekly budget decisions.
Can reports with different windows be compared?
Not meaningfully. A channel judged under seven days last quarter and twenty eight days this quarter will appear to change performance even if nothing real moved. Hold the window constant and label it on every report.