Time-decay attribution is a rule for sharing credit for a conversion that gives most credit to the marketing touchpoints closest in time to the sale or enquiry, and progressively less to those further back. A click the day before someone books counts for far more than an ad they saw three weeks earlier.
How time-decay attribution works
The model uses a half-life. Google’s version, when it was available, used seven days: a touchpoint seven days before the conversion received half the weight of one on the day, a touchpoint fourteen days before received a quarter, and so on. Every interaction inside the attribution window gets some credit, and the shares add up to one conversion.
Take a kitchen fitter in south London. A customer clicks a Facebook ad on 1 March, comes back through organic search on 10 March and clicks a Google search ad on 14 March before requesting a quote. Under last-click attribution the search ad takes all the credit. With a seven-day half-life, the search ad takes about half, the organic visit about a third, and the Facebook ad roughly a seventh.
It sits between single-touch models and the even-handed linear model, and it is a cousin of position-based attribution, which favours the first and last touches instead.
In 2023 Google removed time decay, along with first click, linear and position-based, from both Google Ads and GA4, leaving data-driven attribution and last click. At the time of writing (October 2026), you will find time decay in some other analytics and attribution tools, or you can calculate it yourself from exported path data.
Why it matters
Time decay suits journeys where the decision genuinely builds towards the end: considered purchases made over a few days, offers with deadlines, and services people book after a short spell of comparing. For those, it is a fairer picture than last click, because it acknowledges earlier touches without pretending they mattered as much as the final one.
It undervalues work that starts the journey. For a B2B firm with a three-month sales cycle, or a wedding venue booked a year ahead, the podcast sponsorship or the first LinkedIn post that introduced the brand will look almost worthless. Cutting those channels on the strength of a time-decay report can quietly starve the pipeline the later clicks depend on.
Like any rule-based model, it is an assumption rather than a measurement. It assumes recency equals influence. That may be true of your customers, or it may not.
Common mistakes
- Choosing time decay because it feels fair, without checking how long your customers actually take to convert.
- Comparing figures from tools that use different models and windows, then concluding a channel’s results changed.
- Cutting upper-funnel channels on the model’s say-so without testing what happens when they are paused.
- Forgetting that only tracked touchpoints take part: word of mouth, offline adverts and visits from people who declined cookies never appear.
How to act on it
Look at how long your conversions take first. GA4’s conversion paths report shows the typical number of days and touchpoints before a key event. If most conversions happen within a day or two of the first visit, the choice of model hardly matters. If they spread over weeks, compare a last-click view with a time-weighted one and look at which channels move most between them.
Use any model as one input, then check its suggestions with a controlled test before moving serious budget. Setting up measurement that matches how your customers actually buy is part of how I run performance marketing for UK businesses.