Attribution window

One setting quietly shapes every channel report. A window that is too long flatters slow channels and lets several platforms each claim the same sale, so the total reported conversions exceed the real number. A window that is too short under-credits channels with a long decision cycle and makes them look worse than they are. Cutting a channel that was working, or scaling one that only looked good, wastes budget in either direction.

• Check what each platform uses by default. Defaults differ between platforms and ad types, and platforms report generously on their own behalf.
• Choose one window for comparisons and apply it everywhere, instead of trusting each dashboard's own number.
• Match it to the real buying cycle. If most customers decide within a week, a thirty-day window mostly adds noise.
• Compare platform reports with one neutral source, such as analytics tagged with UTM tags or the CRM.

A buyer clicks an ad on 1 March and buys on 12 March. Whether the ad gets credit depends on the attribution window. With a seven-day click window, the sale falls outside it and the ad receives nothing. With a thirty-day window, the ad is credited. The window is the cut-off that every attribution model works inside.

Most platforms use two windows. A click window covers people who clicked, and a view window covers people who only saw the ad. A view is a much weaker signal than a click, so view windows are normally short.

• Accepting each platform's default window, so the reports cannot be compared.
• Using a long window for a short buying cycle, which adds noise and double-counts sales across platforms.
• Using a short window for a long B2B cycle, which makes slow channels look weak.
• Giving a view-only impression the same weight as a click.
• Changing the window mid-quarter and then reading the change in results as a change in performance.
• Not checking platform totals against the CRM, where the real number of sales sits.