MARK 7397 Spring 2007 Customer Relationship Management A

MARK 7397 Spring 2007 Customer Relationship Management: A Database Approach Class 2 James D. Hess C. T. Bauer Professor of Marketing Science 375 H Melcher Hall jhess@uh. edu 713 743 -4175

Marketing Metrics Traditional Market Share Sales Growth Primary Customer-based Customer Acquisition Popular Customer-based Customer Activity Strategic Customer-based

Traditional and Customer Based Marketing Metrics Traditional Marketing Metrics Market share Sales Growth Primary Customer Based metrics Acquisition rate Acquisition cost Retention rate Survival rate P (Active) Lifetime Duration Win-back rate Popular Customer Based metrics Strategic Customer Based metrics Share of Category Requirement Size of Wallet Share of Wallet Expected Share of Wallet Past Customer Value RFM value Customer Lifetime Value Customer Equity

Primary Customer Based Metrics • Customer Acquisition Measurements – Acquisition rate – Acquisition cost • Customer Activity Measurements – Average interpurchase time (AIT) – Retention rate – Defection rate – Survival rate – P (Active) – Lifetime Duration – Win-back rate

Acquisition Rate • Acquisition defined as first purchase or purchasing in the first predefined period • Acquisition rate (%) = 100*Number of prospects acquired / Number of prospects targeted • Denotes average probability of acquiring a customer from a population • Always calculated for a group of customers • Typically computed on a campaign-by-campaign basis Information source Numerator: From internal records Denominator: Prospect database and/or market research data Evaluation Important metric, but cannot be considered in isolation

Acquisition Cost • Measured in monetary terms • Acquisition cost ($) = Acquisition spending ($) / Number of prospects acquired • Precise values for companies targeting prospects through direct mail • Less precise for broadcasted communication Information source: • • Numerator: from internal records Denominator: from internal records Evaluation: • Difficult to monitor on a customer by customer basis

Average Inter-purchase Time (AIT) • Average Inter-purchase Time of a customer = 1 / Number of purchase incidences from the first purchase till the current time period • Measured in time periods • Information from sales records • Important for industries where customers buy on a frequent basis Information source Sales records Evaluation: Easy to calculate, useful for industries where customers make frequent purchases Firm intervention might be warranted anytime customers fall considerably below their AIT

Retention and Defection • Retention rate (%) = 100* Number of customers in cohort buying in (t)| buying in (t-1) / Number of customers in cohort buying in (t-1) • Avg. retention rate (%) = [1 – (1/Avg. lifetime duration)] • Avg. Defection rate (%) = 1 – Avg. Retention rate Plotting entire series of customers that defect each period, shows variation (or heterogeneity) around the average lifetime duration of 4 years.

Customer Lifetime Duration when the Information is Incomplete Buyer 1 Buyer 2 Buyer 3 Buyer 4 Observation window Buyer 1: complete information Buyer 2 : left-censored Buyer 3: right-censored Buyer 4: left-and-right-censored

Life Table with only right censoring Buyer 1 Buyer 2 Buyer 3 Buyer 4 t Buyer 1: Withdrew late (still active when last observed) Buyer 2 : Withdrew early (still active when last observed) Buyer 3: Terminated late (did not survive past observed date) Buyer 4: Terminated early (did not survive past observed date)

Basic Survival Math S(t) = probability that customer will “survive” until at least time t = 1 -F(t) where F(t) is the traditional “cumulative distribution” f(t) = probability that survival ends at t = -S’(t)=F’(t) 1. 0 S(t) f(t) t 0 S(t 0+t) ----- = Conditional Survival = probability that customer lasts until S(t 0) at least t +t given that they lasted until t 0 0

Hazard Rate and Related Stuff f(t) h(t)= ---- = Hazard Rate= prob that survival ends at t given S(t) that customer makes it to t H(t)=cumulative hazard rate = -ln[S(t)] S(t)=exp[-H(t)] Constant Hazard Rate Model h(t)=h 0, a constant in time H(t)=h 0 t S(t)=exp(-h 0 t) f(t)=h 0 exp(-h 0 t) E[t]= 1/h 0 E[t 0+t | customer made it to t 0] = t 0 +1/h 0 1 h 0 h(t) S(t)=exp(-h 0 t) t

Proportional Hazard Rate Model What if the event varies with customer/situational factors X? h(t) = h. B(t) exp(b. X), where h. B(t) is the baseline hazard rate. * The baseline hazard rate h. B(t) is metaphorically like an “intercept” because when X=0, then exp(b. X)=1. 0 h(t) = h. B(t). If b. X > 0, then exp(b. X)>1. 0, so hazard rates increase above baseline. If b. X < 0, then exp(b. X)<1. 0, so hazard rates decrease below baseline. The coefficients b are chosen in a regression-like fashion, accounting for customer factors and censored data. In SPSS this is done in Survival/Cox Regression. *Why not have h. B(t) b. X? Hazard rates must be positive!

Cox Regression Survival Analysis

Proportional Hazards Assuming Constant Baseline Hazard h(t|Age) = h. B(t) exp(b. X) = 0. 108 exp(-0. 065 Age) E[t 0+t | customer of Age made it to t 0] = t 0 + exp(-b. X)/h 0 = t 0 + exp(0. 065 Age)/0. 108 E[ t | Age made it to t 0] =exp(-b. X)/h 0 =exp(0. 065 Age)/0. 108

Summary • In the absence of individual customer data, companies used to rely on traditional marketing metrics like market share and sales growth • Acquisition measurement metrics measure the customer level success of marketing efforts to acquire new customers • Customer activity metrics track customer activities after the acquisition stage • Lifetime duration is a very important metric in the calculation of the customer lifetime value and is different in contractual and non-contractual situations