The answer to this one admittedly requires some guesswork. But here is the point. It has been proven

The answer to this one admittedly requires some guesswork. But here is the point. It has been proven time after time that data win over perceived expertise, experience, or unaided judgment. Each of these compa- nies uses business analytics, a key form of business intelligence. Business analytics involves the use of computer-based models, statistical analysis, or simulations to test hypotheses about what drives customer loyalty and profitability.

UPS and Harrah’s are examples of two companies that use models to predict when a customer is about to change to another company. At UPS, a phone message is sent to support customers whom the model pre- dicts could switch from UPS. At Harrah’s, the model predicts when customers are reaching their threshold of “too much” gambling losses; a “luck ambassador” is sent to encourage them to take a break and have a free meal, on Harrah’s. Both models are based on extensive data obtained about their customers. Harrah’s system is based on information from a loyalty program, a program in which customers use a swipeable card whenever they use the firm’s services and receive “rewards,” while the company receives vital data about the customers. Capital One uses the same approach to determine the best ways to recruit and retain credit card customers. And the Royal Bank of Canada uses customer data extensively, in combination with ABC costing, to assess customer profitability for each of its banking customers.

Ittner and Larcker, writing in the Harvard Business Review, observe that many companies still make these decisions based on expertise and judgment, but the most successful companies validate their assump- tions about the factors that drive performance in these firms by seeking data and analyzing the data care- fully. Ittner and Larcker present a six step process for validating performance drivers:

Develop a causal model that explains the hypothesized relationship between the performance drivers and desired performance.

Pull together the data that is relevant to test the hypothesized relationships. Loyalty programs such as the one used at Harrah’s are a key source for data.

Turn data into information by using statistical models to test the causal model. Regression analysis is an example of a statistical model that is often used by these companies to test their causal models. It is used at Harrah’s and Capital One, among others (other firms are described in the books by Ayres and Davenport & Harris cited below). Regression analysis is covered in Chapter 8. Continually refine the model

Base actions on findings, that is, trust the statistical model over and above your own judgment and exper- tise. This is like trusting your GPS to tell you where you are rather than to go by your own guesswork.

Assess outcomes by an ongoing review of the outcomes of your actions, to determine if the model needs revision, the competitive environment has changed, and so on.