A famous quote by Sam Walton
Determining the value of a customer is a holy grail in CRM Anaytics. Organizations adopt different methods to rank a customer in terms of value they bring to the organization. Marketing efforts are designed an spent on customers based on value rank.
Two popular methods of determining customer value are
- RFM (Recency, Frequency, Monetary)
- This is a simple and easy to implement solution
- Performs ranking only and parameters can be dynamically adjusted
- It requires basic data about a customer’s history of transactions with the organization
- CLTV (Customer Life Time Value)
- Complex approach and implementing takes more time
- Requires more data for analysis to run
- Mathematical model does not provide room for changes
- Tries to predict future prospects of a customer
Organizations adopt RFM based analysis for customer ranking first and then graduate to CLTV based analysis. RFM analysis deals with three parameters and all three parameters are independent of each other, but are tied together by time context. A common time context should be used or else the final rank will be skewed.
Firstly, historical customer data is used to determine the different scores for R, F & M. Method for determining scores depends on the business policies, which are adjusted based on different parameters. In the example above, assume that a three value scale is implemented i.e. Low = 1, Medium = 2 and High = 3
- Recency: If customer had purchased in last 30 days, then “High” or else if purchases were made in last 90 days then “Medium” or else “Low”. In this case, score for Recency is 2.
- Frequency: If the customer had purchased in last 30 days 5 or more times, then “High” or else if purchases were made in last 90 days 3 or more times then “Medium” or else “Low”. In this case, score for Frequency is 2.
- Monetary: If the customer had purchased in last 30 days value of $5000 or more, then “High” or else if purchases were made in last 90 days value of $10000 or more times then “Medium” or else “Low”. In this case, score for Monetary is 2.
The above example illustrates a simple scoring model for RFM analysis. Based on the scores arrived, a simple rank or weighted rank could be obtained as the product of three parameters as follows.
Simple Rank = 2 * 2 * 2 = 6 out of 9
Weighted Rank = [(50% * 2) + (30% * 2) + (20% * 2)]/100 = 66%
Rank thus determined is used for further analysis and the process is tweaked over period of time whenever there are changes to internal and external factors. In real-time scenarios the scoring models are complex and when used in conjunction with predictive analysis can really give deep insight into customer behavior.