bachmannpatrick / CLVTools

R-Package for estimating CLV
54 stars 14 forks source link

Why average monetary value instead of total monetary value? #207

Closed SSMK-wq closed 2 years ago

SSMK-wq commented 2 years ago

I read about Recency, Frequency and Monetary technique and I see that a customer worth is determined by his total revenue that company achieved from a specific customer.

But I see that for RFM analysis the package, uses Average revenue. Is there any reason why you consider Average revenue only? If a new customer comes with 2 transactions places huge order for 100K and 200K, then he contributed 300K to the company and could be classified as "New but promising" or "New but Heavy spender" etc.

But doesn't taking average normalize everyone on the equal scale? So, then monetary value doesn't become useful metric to segment customers. Instead we have to use only Recency and Frequency (because they have raw values).

Is there any reason why you think average revenue is better than Total revenue?

pschil commented 2 years ago

Im not sure if I can follow entirely. This package does not do RFM analysis. I recommend reading the vignette accompanying the package which can also be found here: Link

The transaction models (Pareto/NBD, BG/NBD, etc, all models except the Gamma/Gamma) predict how many more (repeat) transactions each customer will make in the prediction period: CET. The expression DERT which is available for some models, is the number of transactions until Infinity, multiplied with a discount factor (time value of money). See here for more details: Link.

The Gamma/Gamma spending model predicts how much a customer will on average transact for each of these predicted transactions. Output is: average monetary value per transaction

CET * (average monetary value per transaction) = The expected monetary value of each customer in the prediction period ("how much the customer will spend in total during the prediction period")

DERT* (average monetary value per transaction)= Customer Lifetime Value (CLV), how much this customer is worth over his entire lifetime (until Infinity)

Both values give you an impression of "how much more value" is left in your customers and you can then do segmenting based on either of them. The resulting values actually take into account all components of RFM. Internally, the models do only require RFM to work, part of what makes them so elegant.