Our recommended attribution model is Multi-Touch Time decay. This weights the fractional credit along the timeline of exposure; whereas, the ad exposure (within the attribution window) that occured weeks ago receives the least amount of credit & the one that happened most recently gets the most credit.
The Decay Formula (7-day Attribution Window Example):
Decay Factor=2−(time_window/7)
Where:
- time_window represents the time (in days) between the ad exposure (i.e., when the episode with the ad was listened to) and the conversion event.
- 2 is the base of the exponential decay, meaning the credit reduces by half over a set period (in this case, a 7-day window).
This formula is applied to each touchpoint, and the credit for each touchpoint is calculated based on its time distance from the conversion.
How it works:
- Exponential Decay: Exponential decay means that the credit given to each touchpoint decreases as the time between the touchpoint and the conversion increases. The decay factor halves approximately every 7 days, making recent touchpoints more valuable than earlier ones.
- Relative Credit Assignment: For each touchpoint, we calculate the time window from the conversion and apply the decay formula. The most recent touchpoint will have the smallest time window (e.g., 0 days), and thus its decay factor will be closer to 1, giving it the most credit. Touchpoints further from the conversion will have larger time windows and lower decay factors, reducing their assigned credit.
Example:
Let's say a listener hears ads on three different episodes before making a conversion. The conversion happens after hearing the last episode.
- Touchpoint 1 (last episode): This episode is 1 day before the conversion.
Decay Factor=2−(71)≈0.90
- Touchpoint 2 (second last episode): This episode is 4 days before the conversion.
Decay Factor=2−(74)≈0.65
- Touchpoint 3 (third last episode): This episode is 10 days before the conversion.
Decay Factor=2−(710)≈0.37
Assigning Credit:
Once we calculate the decay factors for each touchpoint, we normalize them to sum up to 100%. So, in this case:
- The most recent episode gets the highest credit (with a decay factor of ~0.90).
- The second last episode gets the second-highest credit (with a decay factor of ~0.65).
- The third last episode gets the least credit (with a decay factor of ~0.37).
To convert the decay factors into percentages for total credit assignment:
- Add up the decay factors: 0.90+0.65+0.37=1.92
- Normalize each touchpoint by dividing its decay factor by the total:
- Last episode: 1.920.90×100≈47%
- Second last episode: 1.920.65×100≈34%
- Third last episode: 1.920.37×100≈19%
Thus, the most recent episode gets 47% credit, the second-to-last gets 34%, and the third-to-last gets 19%.