Time-Decay Attribution
A multi-touch attribution model that assigns increasing credit to touchpoints closer to the conversion event, based on the assumption that more recent interactions had greater influence on the purchase decision.
Also known as: time decay model, recency-weighted attribution
Why time-decay attribution matters
Time-decay attribution strikes a practical balance between giving credit to all touchpoints and recognizing that recency matters. A product demo yesterday likely influenced the purchase more than a blog post three months ago, even if both played a role.
This model is particularly well-suited for businesses with longer sales cycles where many touchpoints accumulate over time. In a 90-day enterprise sales process, the early-stage awareness touchpoints receive some credit but are appropriately weighted less than the late-stage evaluation and decision touchpoints.
Time-decay also produces results that align with how most marketers intuitively think about attribution. The gradual reduction in credit over time feels more natural than the sharp cutoffs of first-touch or last-touch, making it easier to get organizational buy-in for the model's results.
Stay ahead in analytics and growth
The most important reads on measurement, retention and revenue, in five minutes a week, in a free email. Already read by 10,000+ operators and founders.
No spam. Unsubscribe in one click.
How to calculate time-decay attribution
Time-decay models typically use a half-life parameter. If the half-life is 7 days, a touchpoint 7 days before conversion receives 50% of the credit of the most recent touchpoint, a touchpoint 14 days before receives 25%, and so on. The credits are then normalized to sum to 100%. The decay function is: credit = 2^(-time_since_touchpoint / half_life), normalized across all touchpoints.
Time-Decay Attribution examples
A consumer electronics retailer uses a 7-day half-life time-decay model. Retargeting ads and abandoned cart emails receive the most credit (being closest to conversion), while social media awareness campaigns receive less but measurable credit. This balanced view prevents the team from cutting social entirely.
An enterprise security company uses a 21-day half-life for their 90-day average sales cycle. The model shows that late-stage touchpoints (demo, pricing page, ROI calculator) receive the most credit, but early-stage analyst reports and industry events still receive 15-20% of total credit, validating continued investment.
How to Track in Kissmetrics
Kissmetrics does not offer a time-decay model. The Performance report credits each conversion by first touch or by last touch, and last touch is the closest to time decay: it gives all the credit to the most recent campaign. To model decay yourself, export each person's campaign visits with their timestamps and apply the half-life you choose in a spreadsheet or BI tool.
Common Mistakes
- -Using a half-life that does not match your actual sales cycle, which over-credits or under-credits early touchpoints
- -Applying time-decay to short journeys with only 2-3 touchpoints where the decay effect is negligible
- -Not testing different half-life settings to understand how sensitive your channel rankings are to this parameter
- -Forgetting that time-decay still undervalues awareness channels that are structurally positioned early in the journey
Pro Tips
- +Test half-life values of 7, 14, and 30 days and see how channel rankings shift - stable rankings across settings indicate robust attribution
- +Use a shorter half-life for ecommerce (7-14 days) and a longer one for B2B SaaS (14-30 days)
- +Combine time-decay with position-based attribution for a hybrid model that values both recency and position
- +Use time-decay for operational channel management and incrementality testing for strategic budget decisions
Related Terms
Decay Model
An attribution approach that assigns decreasing credit to touchpoints based on their distance in time from the conversion event, giving more weight to recent interactions and less to earlier ones.
Multi-Touch Attribution
An attribution approach that distributes conversion credit across multiple touchpoints in a customer's journey rather than assigning all credit to a single interaction, reflecting the reality that most conversions involve multiple marketing influences.
Linear Attribution
A multi-touch attribution model that distributes conversion credit equally across every touchpoint in the customer journey, giving the same weight to the first, middle, and last interactions.
Position-Based Attribution
A multi-touch attribution model that assigns the most credit to the first and last touchpoints in the customer journey (typically 40% each) while distributing the remaining credit equally among middle interactions.
Attribution Model
A set of rules or algorithms that determine how credit for conversions and revenue is assigned to the marketing touchpoints in a customer's journey, shaping how channel ROI is measured and budget is allocated.
Further Reading
Cost Per Acquisition for SaaS: Why You Need Customer Analytics to Get It Right
Learn how to calculate true CPA for SaaS including multi-touch attribution, channel-level analysis, and the relationship between CPA and customer lifetime value.
UTM Parameters: A Taxonomy That Survives to Revenue
A UTM has to survive a redirect, a return visit, a device switch and several weeks before it reaches a payment. A naming taxonomy built for that journey.
How to Build a Marketing Analytics Stack That Proves ROI
A marketing analytics stack proves ROI only if every campaign resolves to revenue. Which tools to connect, which attribution to run, how to read it.
Marketing ROI: How to Measure What Matters and Prove Your Impact
Marketing ROI is revenue attributed to marketing minus its cost, over the cost. The formula, the attribution it depends on, and how to prove campaign impact.
Try it on your own data for free.
First touch and last touch on events tied to people, with signups and revenue per person for each campaign, plus a report for visitors from AI assistants. Both come with the Growth plan.