Time-Decay Attribution

Time-decay attribution is a MULTI-TOUCH-ATTRIBUTION model that distributes credit for a conversion across all the touchpoints in the journey, but weights them by recency: touchpoints closer in time to the conversion get more credit, and earlier ones get progressively less (the credit 'decays' as you go back in time). It's a middle ground between LAST-CLICK-ATTRIBUTION (all credit to the final touch) and FIRST-CLICK (all to the first), spreading credit across the journey while still emphasizing the touches nearest the conversion — a reasonable compromise with a built-in recency bias.

How it works and where it fits: time-decay uses a decay function (often a half-life — a touch X days before conversion gets half the credit of one at conversion, and so on) to weight touchpoints, so the final touches before conversion get the most credit and the earliest get the least, with everything in between getting a recency-weighted share. Among the attribution models it sits in the middle: last-click (all credit to the final touch — maximum recency bias, ignores the journey), first-click (all to the first — ignores the closing), linear (equal credit to every touch — no recency weighting), time-decay (recency-weighted credit across the journey — the recency compromise), position-based/U-shaped (extra credit to first and last, less to the middle — emphasizing discovery and closing), and DATA-DRIVEN-ATTRIBUTION (algorithmic credit — more sophisticated). Time-decay's logic and its appeal: the rationale is that touchpoints closer to the conversion were 'more influential' in driving it (the recency-matters intuition), and it's a reasonable improvement over last-click (it at least spreads credit across the journey rather than giving everything to the final touch) while being simpler than data-driven models. The built-in bias and the honest caveat this entry must center: time-decay has a RECENCY bias by design — it systematically gives more credit to the touchpoints nearest the conversion (the demand-HARVESTING channels — branded search, retargeting, the final email) and less to the early touchpoints (the demand-GENERATING channels — the awareness and discovery that started the journey), so like all correlational attribution models it can under-credit the upper-funnel work that created the demand the recent touches harvested (the INCREMENTALITY-VS-ATTRIBUTION concern, milder than last-click's but still present — time-decay assumes recency equals influence, which isn't always true, since the early awareness touch may have been the decisive one). The framing: time-decay attribution is the recency-weighted multi-touch model — crediting touchpoints more the closer they are to conversion — a reasonable middle ground that improves on last-click by spreading credit across the journey while still emphasizing the closing touches, with a built-in recency bias that under-credits early demand creation; the discipline is using it as one correlational attribution view (better than last-click, simpler than data-driven) while remembering its recency bias under-credits the upper funnel, and validating with incrementality rather than trusting any attribution model — including time-decay — as causal truth, since time-decay's recency weighting is a reasonable heuristic, not a measurement of what each touch actually caused.

Time-decay attribution matters as a multi-touch model option — a reasonable middle ground for crediting the journey while emphasizing the closing touches, better than last-click (which ignores the journey) and simpler than data-driven models, useful where a recency-weighted view of the multi-touch journey fits. It matters with awareness of its built-in recency bias (it under-credits the early demand-generating touches and over-credits the recent demand-harvesting ones) and as one correlational view, not causal truth. The discipline is using time-decay as one attribution lens (recency-weighted multi-touch) while remembering its recency bias under-credits the upper funnel, reading it alongside other models and the funnel position of credit, and validating budget decisions with incrementality rather than trusting time-decay (or any attribution model) as a measurement of what each touch actually caused — treating its recency weighting as a reasonable heuristic for crediting the journey, not the causal answer.

Time-decay attribution emerged among the multi-touch attribution models as marketers sought middle ground between last-click and first-click - crediting the whole journey while emphasizing the touches nearest conversion via a decay function; it remained a reasonable correlational heuristic with a built-in recency bias, superseded for causal questions by incrementality testing.

Etymology: source.

Search interest for this term over the last five years:

View interest-over-time on Google Trends →

Areas of marketing where time-decay attribution is a core concern: