time decay attribution

Time decay sits in the multi-touch family: rather than handing a full conversion to one interaction, it splits credit across every touchpoint in the path, then weights that split by recency. Google’s version used a 7-day half-life inside a 30-day lookback window.

Worked example (before normalizing to 100%):

Touchpoint on conversion day:   weight 1
Touchpoint 7 days before:       weight 1/2
Touchpoint 14 days before:      weight 1/4

The logic fits short sales cycles and promotions, where a click yesterday plausibly mattered more than one from three weeks ago.

Where it stands in 2026

Google removed time decay (along with first-click, linear, and position-based) from Google Analytics 4 in November 2023 and retired it in Google Ads as well. GA4 now offers data-driven, paid-and-organic last click, and Google-paid-channels last click; conversions that used time decay were migrated to data-driven. The model still exists in other analytics tools and stays a useful reference point when you compare how different models split credit.

The catch is recency bias. Time decay systematically underweights the discovery moment, which for organic search is often the first informational or branded query weeks before the purchase. Read it knowing it flatters closing channels and discounts the ones that seed demand.