WebmasterID logoWebmasterID
Attribution models

Time-decay attribution: recent touches weigh more

Time-decay attribution weights touchpoints by recency: the closer a touch is to the conversion, the more credit it earns, usually following an exponential decay with a configurable half-life. It is a compromise between last-click and linear, but its recency bias under-credits the early demand-creating touches.

Verified against primary sources

What this means

Time-decay applies a decay curve so touches near the conversion earn the most credit and older touches earn progressively less. A 'half-life' parameter sets how fast credit fades — a seven-day half-life means a touch a week earlier is worth roughly half a touch at conversion.

Recency is an assumption

The model assumes recent touches are more influential. That fits some journeys (an impulse purchase) and badly distorts others (a long B2B evaluation where the first webinar created all the demand). Changing the half-life changes the credit split, so two teams using 'time-decay' can disagree sharply.

It is a useful middle ground, but the half-life is a judgement call, not a measurement.

How it appears in analytics and logs

Time-decay output reflects recency as much as influence. A channel that consistently appears late in paths will look strong; early discovery channels will look weak by design.

Diagnostic use case

Use time-decay when later touches plausibly matter more (short consideration cycles), while remembering it built-in penalises early awareness work.

What WebmasterID can help detect

WebmasterID keeps weighting choices explicit and confidence-labelled, so a recency-weighted view is read as one assumption among several, not as truth.

Common mistakes

Privacy and accuracy notes

Time-decay needs only the timestamps of one site's own touchpoints relative to the conversion. No cross-site identity is required.

Related pages

Sources and verification notes

Last reviewed 2026-06-24. Facts are checked against primary/official sources where available; uncertain specifics are marked “Data not yet verified” rather than guessed.