Calculating Recency with Auto-Updating Sales Data

Hi lbendlin  ,

Thank you for the example. I calculate recency as the time from the last transaction to the last date of analysis (in this case, May 31).

So, for Customer A:

If the transaction on May 1st is the last one in May, the recency is 31−1=30 days.

For Customer B:

Similarly, if the transaction on May 20th is the last one in May, the recency is 31−20=11 days.

From what I understand, you are suggesting calculating the recency as the number of days between the penultimate purchase and the first purchase of the selected period. However, this would not be a correct calculation of recency and could lead to misleading results.

Example:

Customer C:
First purchase of the period: May 15th
Penultimate purchase: January 1st

Using your suggested method, the recency would be 135 days, whereas calculating it as I suggest, it would be 16 days.

With the first calculation method, I would think that this is a dormant customer who needs to be reactivated, while with the second calculation method, this is a customer I consider active, which they actually are.

Could you please confirm if my interpretation of your suggestion is correct?

Thank you for your help!