Managing Customers for Profit

1. Managing Customers for Profit V. Kumar Chapter – 3 Customer Selection Metrics Instructor’s Presentation Slides

2. Relevant Issues

3. Types of Customer Selection Methods • Traditional Metrics • Recency-Frequency-Monetary Value (RFM) • Frequently used in mail-order companies • Share of Wallet (SOW) – • Used in High Tech firms • Past Customer Value (PCV) • Used in the financial services industry • CLV is an advanced metric that is widely gaining popularity across all industries

4. Recency-Frequency-Monetary Value • Used in Mail order and catalog industries (71%) • This technique uses past customer information about Recency, Frequency, and Monetary Value to evaluate and predict customer behavior and customer value and is defined as follows: 1. Recency is a measure for the time elapsed since a customer last placed an order with the company. 2. Frequency is a measure of how often a customer orders from the company in a certain time period. 3. Monetary value is a measure of the amount that a customer spends on an average transaction.

5. RFM: An Example • ABC Sports ware can contact one of its two customers (they are REALLY good customers…) • ABC has purchase behavior data for each of these two customers over a period of five months (Jan-May) Dollar Amount Spent by Each Customer Purchase frequency of ABC’s Customers

6. RFM: An Example (solution)

7. Share-of-Wallet (SOW) • Share-of-wallet (SOW) indicates the degree to which a customer meets his or her needs in the category with a focal brand or firm. • SOW is used widely in retail businesses such as supermarkets and in financial companies mainly to identify whether or not consumers are loyal to a specific store or whether they shop around at different stores. • This metric provides B2B companies with an idea of what portion of the marketing budget is being spent with that firm. • SOW can be estimated either at the individual level or at the aggregate level, but as a customer scoring metric it used almost exclusively at the customer level

8. Share-of-Wallet (SOW) • Individual Share-of-Wallet= where, S = sales to the focal customer, j = firm, = sum of the value of sales made by all firms that sell products to a certain buyer.

9. SOW Example Table 3.1: Dollar Amount Spent by Each Customer Table 3.2: Purchase Frequency of the Customers of ABC Sportswear

10. SOW Solution

11. Past Customer Value Past customer value of a customer = where, i = number representing the customer, r = applicable discount rate (for example, 15% per annum or 1.25% per month), T = number of time periods prior to the current period when the purchase was made, GCit= gross contribution of transaction of the ith customer in time period t. Table 3.6

12. PCV Solution Table 3.7: PCV of the Customers • The gross contribution made by each customer is given in Table 3.6. The PCV score for the two customers (from the original ABC Sportswear example) is calculated based on the formula given in equation (2) (assuming the discount rate as 1.25% per month). • Based on the PCV scores, ABC Sportswear decides that Customer 1 has the greater potential for profitability since the PCV score for Customer 1 ($180) is higher than for Customer 2 ($117). Thus, in this case the ABC Sportswear Company chooses Customer 1 as the ideal customer for its current marketing campaign.

13. Shortcomings of Metrics

14. Customer Lifetime Value (CLV) • Unlike other traditional measures that include only past contributions to profit, the merit of CLV rests on the fact that it is a forward-looking metric. • CLV assists marketers to adopt appropriate marketing activities today, in order to increase future profitability • The computation can also be used to include prospects, not just current customers as used by the RFM, SOW, and PCV metrics • The computation can also be used to include prospects, not just current customers as used by the RFM, SOW, and PCV metrics • CLV also manages to score over other metrics by adopting a customer-centric approach instead of a product-centric one, as the driver of profitability.

15. - Gross Contribution Margin Marketing Costs - Accumulated Margin Acquisition Costs Adjusted for Present Value - Recurring Revenues Recurring Costs Customer Lifetime Value Expected Number of Purchases over next 3 years X Net Margin Introduction to CLV Figure 3.1 CLV Measurement Approach

16. CLV Decisions

17. Measuring CLV Aggregate Approach for Measuring CLV CE = customer equity of the customer base in U.S. dollars (sum of individual lifetime values), CM = average contribution margin in time period t, (after taking into account marketing costs) d= discount rate, i = customer index, t = time period, T = the number of time-periods for which CE is being estimated. Cohort Approach for Measuring CLV r = rate of retention, d = discount rate or the cost of capital for the firm, t = time period, T = the number of time periods considered for estimating CE, GC = the average gross contribution, M = marketing cost per customer, A = the average acquisition cost per customer.

18. Case Study: CLV • The purchase behavior of a cohort of 100 customers in a financial services firm, NSE Inc., was observed over a period of three months starting January 2005. • The average gross contribution from this group of customers in the next month (April 2005) is projected to be $500. • The marketing cost and acquisition cost per customer is $45 and $60, respectively. • The discount rate is 15% per year or 1.25% per month. • The retention rate for this cohort is 0.75. • Assuming that the discount rate, retention rate and gross contribution are constant over time, the average CLV of a customer belonging to this cohort over the next three time periods is given as:

19. Case Study: CLV (cont’d) • However, average CLV has limited use as a metric for allocation of resources across customers. • The average CLV metric does not capture customer-level variations in CLV, which is the basis for developing customer-specific strategies. • Calculating the aggregate CLV does not allow corrective measures to be implemented at the segment-level or individual customer level. • Hence, it is necessary to calculate the CLV of individual customers in order to design individual-level strategies.

20. (Future contribution margin it – Future cost it) (1+d)t Measuring CLV: An Individual Approach • The calculation of CLV includes determining the future contribution margin and future costs, both of which are adjusted for the time value of money i = customer index, t = time index, T = the number of time periods considered for estimating CLV, d = discount rate.

21. P( Active) • P(Active) - the probability that the customer continues to be active in a subsequent time-period • Calculation of this probability at an individual level is essential for CLV calculation at an individual level as each customer is likely to have different purchase patterns and inactive periods. Figure 3.2: Variation in Inter-purchase time Customer 1 Customer 2 Month 1 Month 8 Month 12 Source:Reinartz, W.J., & V. Kumar (2002), “The Mismanagement of Customer Loyalty,” Harvard Business Review, 80(7), 86.

22. P(Active) cont’d P (Active) = (T/N)n n = the number of purchases in the observation period, T = the time period of the most recent purchase, and N = the current time period for which P (Active) needs to be determined. In Figure 3.2, the “star” indicates a purchase made by a customer. Therefore, for Customers 1 and 2, the P (Active) for month 12 would be: P (Active) for Customer 1 in Month 12 = (8/12)4 = 0.197, where n = 4. P (Active) for Customer 2 in Month 12 = (8/12)2 = 0.444, where n = 2.

23. Average Monthly Gross Contribution (AMGC) • The firms ascertain the average gross contribution margin (AMGC) by deducting the average cost of goods sold from the average monthly revenue from a customer. • This is calculated based on the customer’s past purchases. This is obtained for all customers i and for the time-period t for which the lifetime value is being estimated. • To arrive at the present value of the future contribution, the AMGC of the customers is adjusted with a discount rate d, for the number of time-periods n.

24. Net Present Value (NPV) (8) NPV of EGC it= AMGCit = average gross contribution margin in period t based on all prior purchases, i = customer index, t = the period for which NPV is being estimated, x = the future time-period, n = the number of periods beyond t, d = discount rate, P (Active) in = the probability that customer i is active in period n.

25. Case Study • Consider the spending pattern of the two customers of ABC Sportswear, from the previous sections. • Refer to Table 3.6 for the spending patterns of the two customers over a five month period (Jan-May). • Find AMGC and NPV.

26. Calculating CLV • To calculate the lifetime value of a customer, the acquisition (A) and the marketing costs (M) incurred at future time periods have to be deducted from the NPV of EGC of a customer. • The marketing costs at future time periods should be discounted with the appropriate discount rate (d), in order to arrive at the present value of these costs. • The discounted marketing costs (M) and the acquisition cost (A) are then subtracted from the NPV of EGC to arrive at the CLV of a customer.

27. Calculating CLV (Contd.) CLV of customer i = AMGCit = average gross contribution margin in period t based on all prior purchases, i = customer index, t = the period for which NPV is being estimated, x = the future time-period, n = the number of periods beyond t, d = discount rate, P (Active) in = the probability that customer i is active in period n. M = the marketing costs of the firm A = the acquisition costs of the firm

28. CLV Components

29. Advanced CLV Model • A “lost-for-good” approach assumes that once customers are gone, they are gone forever. This often underestimates CLV. A better approach is to use an “always-a-share” approach. • A more advanced model for calculating CLV models the inter-purchase time (quantified by frequency) instead of P (Active). • The inter-purchase time is modeled by fitting a distribution over the past inter-purchase behavior of the customer and by taking the expectation value of this distribution. • Using frequency instead of P (Active) accounts for the customers who are dormant for a particular period of time (as it frequently occurs in customer purchases like automobiles or computers) and who then come back to the firm. • This provides a more realistic and robust prediction of the purchase behavior of customers.

30. Advanced CLV Model (cont’d) • Three parameters to predict • Future customer activity (frequency) • Future marketing costs (MC) • Gross contribution margin from each customer (GC)

31. (12) Purchase Frequency Model Drivers of PFM • Number of product purchase upgrades • Cross-buying behavior of customers (across product categories) • Ratio of number of customer-initiated contacts to total contacts (customer initiated and supplier initiated) • Product return behavior • Frequency of web-based contacts • Frequency of customer contacts (in-person, direct mail and telephone) by the firm • Average time between two customer contacts Drivers of the gross contribution margin (GC) model : • Customer’s contribution margin from the previous year • Total number of customer contacts across all channels • Total quantity purchased across all product categories Two methods can be used to forecast the future marketing cost (MCi,l,m): • The first method assumes that the past cost will continue in the future, if there is not much change in marketing costs at the customer level over the years. • The second method considers the future marketing cost as a function of current purchase activity and current marketing cost.

32. (12) PV of Marketing Cost PV of Gross Contribution Purchase Frequency Model CLV= Customer Lifetime Value GCi,t= Gross contribution from customer i in purchase occasion t MCi,l,mrefers to Marketing cost, for customer i in communication channel m in time period l. where, MCi,l,m= ci,m,l (unit marketing cost) * xi,m,l (number of contacts) frequencyi = 12/expinti (where, expinti = expected inter purchase time for customer i) r is the discount rate for money n is the number of years to forecast, and Ti is number of purchases made by distributor i, until the end of the planning period

33. PV of Gross Contribution Case Study Time period (l) = 2 years Frequency of purchase = 2 purchases per year Unit cost of telesales = $30 No. of predicted contacts through telesales = 25 in year 1 & 20 in year 2 Unit cost of salesperson = $600 No. of predicted sales contacts by salesperson = 10 in year 1 & 15 in year 2 Discount Rate (r) = 15% The CLV score can be calculated for the above data as follows:

34. Regularity Frequency RFM Tenure CLV r= - 0.09 r= 0.17 r= 0.19 r= 0.44 N 172,688 470,932 470,932 470,932 Step 1: Identifying the Drivers of Loyalty • The retailer used several measures to identify loyal customers: • Regularity of Purchase • Frequency of Purchase • Tenure Question: Do these measures of loyalty drive profitability? Result: Except for tenure, the traditional metrics of loyalty showed poor correlation with loyalty.

35. Step 2: Measuring CLV The lifetime value is computed for each customer using this formula: Where: GCi,t = Gross contribution from customer i in purchase occasion t ci,m,l = unit marketing cost, for customer i in channel m in time period l xi,m,l = number of contacts to customer i in channel m in time period l frequencyi = 12/expinti, expinti = expected inter purchase time for customer i r = the discount rate for money n = is the number of years to forecast Ti= total number of purchases made by customer i

36. Step 3: Scoring & Segmenting the Customers Customers are rank-ordered into deciles and segmented based on the distribution of CLV across the deciles High CLV Medium CLV Low CLV

37. Step 4: Identifying the drivers of CLV B X CLV Score ($) Amount of Returns ($)

38. -5% 0% 5% 10% 15% 20% 25% 30% 35% % Change in CLV Step 5: Interpreting The Impact of The Drivers A 15% increase in cross-channel spending by customers in the top 2 CLV deciles results in 31% increase in their CLV for PRC stores. Similar interpretation holds for the remaining variables illustrated below. Lifetime Returns -3% $ Spent on Product B 4% Cross-Selling 11% $ Spent on Product A 12% Tenure 14% $ Spent in other channels 31%

39. Step 6: Profile Analyses Develop profile analyses for the High and Low CLV Customers Typical High CLV Customer Typical Low CLV Customer http://www.drvkumar.com/books/book_mcp.html ©Dr. V. Kumar

40. When to stop chasing??? The important factor in deciding when and which customers to let go of is the measurement of profitable customer lifetime duration • Decision Rule: • If NPV of Expected contribution margin is less than cost of mailing, then the firm would decide to terminate the relationship • Using the above decision rule, we can establish for every customer at what point he is subjected to the proposed termination policy

41. End of Chapter 3