jim.shamlin.com

7: Using the Database to Separate Good and Bad Customers

Given that businesses have been computerized for a few decades, there is a wealth of data available in company databases that can provide considerable insight into customer relationships, if only it could be identified, extracted, and analyzed in a meaningful way.

The key problem is that information in databases is often reserved and controlled for the functional needs of a given system, and that the data is unavailable to those managers who could benefit from having access. Most commonly, it is locked up in the IT department, which uses security/privacy concerns as a reason to deny access to others in the organization - hence it is a political and cultural challenge to unfreeze this data and provide it to those who can benefit from having it. And since it is ultimately the company that stands to benefit, it is an effort well worth undertaking.

WHAT MARKETERS KNOW ABOUT CUSTOMERS: TYPES OF CUSTOMER DATA

The author lists some of the sources where data is collected and stored:

While there has been a movement toward providing data solutions on the enterprise level, there remains the problem of business units and subsidiaries maintaining their own information and resources that are not available to the rest of the business. As a result, there are dozens to hundreds of data sources within an organization that are redundant and uncoordinated.

(EN: Coordinating the data in an organization is a task for the IT department - so I'll skip the information on this, except to note that marketers should be eager to support such initiatives, but may need to cope with fragmented and dispersed systems in the meantime.)

The author distinguishes between measured and implied data: Measured data is concrete information about past actions, whereas implied data may either be a projection of an expected future state, or conclusions about the past that are derived from measured data.

The author also separates observable from projectable, which differentiates data that can be rigidly measured (how many times a customer booked a window seat versus an aisle seat) versus that which is assumed (because the customer books window seats more often, we believe they prefer a window seat and will book them in the future).

LEVERAGING CUSTOMER INFORMATION: A FRAMEWORK

The author lists a number of sources from which customer information is obtained, including both operational sources (transaction data), ancillary information collected about the customer (marketing data), as well as any information gained from third-party sources.

Ideally, this information is aggregated in a central data warehouse, rather than being stored in a multitude of systems, which should contain complete information about each customer, including present data and historical data, that can be accessed by the various functional applications as well as being aggregated for reporting and analysis.

From a marketing perspective, the value of this information is in its ability to contribute to a model or profile of its primary market segments, which can be analyzed to predict the reaction to future business activities.

Even though the cost of computer hardware has decreased significantly, there still remains the decision of data retention: how much information about the customer is retained in information systems has a significant cost impact, and may have a performance impact. No single answer suits all firms.

Primarily, a firm must consider the value of the individual customer. A manufacturer of a cheap commodity good, such as toothpaste, has a large number of customers, each of whom represents a small amount of value to the firm, and the relationship with the customer is fairly weak, so maintaining data is not of much value. However, an airline will find that each ticket purchased carries a significant gross margin, and there si value in capturing and storing information as well as cultivating an ongoing relationship with a customer to retain their future business.

Another key variable is the degree to which data is used in marketing programs - which follows the general principle that any data that is used is of some value. Taking the same two examples, a company that sells toothpaste does not need to tailor its marketing campaigns to the individual customer, but deals with broad market segments with similar needs and behaviors, hence its need for data is low; whereas an airline maintains a close relationship with its "frequent flyer" customers and requires granular information about their travel patterns to extend offers that a given customer will find appealing.

(EN: My sense is this is implicit in the difference in B2B versus B2C marketing. In the example of the toothpaste manufacturer may have little interest in the consumer as an individual, but would probably benefit from collecting data and cultivating relationships with wholesalers and retailers who buy in volume.)

USING AND ANALYZING CUSTOMER DATA

The author proposes to explain some of the analytical tools and techniques that "turn raw data into compelling customer insights"

(EN: The author completely botches the description of statistical analysis, and I fear the notes will do more to confuse than clarify the concepts, so I'm skipping that detail.)

Of importance is that aggregation (combining users into groups) and separation (dividing large groups into smaller ones) should not be an arbitrary matter - one should not arbitrarily separate customers by age into 10-year segments (10-19, 20-29, 30-39, etc.) unless there is statistical evidence that the behavior of these age groups is also similar.

(EN: A bit more detail on this notion might serve to clarify: arbitrary age groups are especially problematic with younger consumers. The 10-19 group includes several groups whose behaviors are significantly different in their behaviors, yet an arbitrary age grouping lumps them together and treats a 10-year-old child and a 19-year-old college student as being the same in their behaviors, tastes, attitudes, incomes, etc.)

The author goes on about this for a while, and waxes a bit preachy at times, but his point is well-taken: given the capacity and capability of computer technology that exists today, and the wealth of data that is being collected and stored, there simply is no excuse for not making good use of it to understand your customers better. The fact that many companies simply don't bother makes this a fertile ground for developing competitive advantage.

VALUING CUSTOMERS AND PROSPECTS

Especially in relationship marketing, significant resources are devoted to courting the customer - and all customers are not of equal value. As such, a company should seek to invest the most effort in maintaining the relationship with its highest-value customers - which begs the question: how much is a given customer worth?

Following the Pareto principle, it's likely that 20% of the customers of a firm account for 80% of its revenues or profits. In terms of finances, then, the value of a customer (or group of customers) is based on the income flow the customer creates for the company through their ongoing patronage.

To that end, the author identifies four characteristics that should be considered in determining the value of a customer or segment:

(EN: I find this a bit questionable, For example, penetration is probably useful for comparing customers of competing firms, but not for sorting the customers of a single firm, and the deduction of marketing cost from revenue effectively "blames" the customer for something over which he has no control - if the company frivolously spends money on bad marketing, this makes the customer seem less valuable to the firm - though customer-driven costs such as number of service calls or product returns certainly should be factored into their overall value, as some customers cost more than they bring in.)

(EN: While I haven't quite worked it out, it would seem that there are two statistics to be determined: the present value of a customer based on purchase frequency and margin, versus the potential value of that customer if you could gain 100% share of wallet, which are the "as is" you seek to preserve and the "to be" that you hope to achieve. This may require further consideration.)

There's also the notion of product-line segmentation. The author provides an example of a department store in which the "best" customers in terms of total annual revenue per customer purchase sporadically, in a small number of categories, whereas the below-average customer in terms of revenue is a more frequent shopper in other product categories. Another example is the credit-card industry, in which the customers who generate the most revenue for the firm are those who make fewer purchases, but carry more of a balance from month to month, that the customers who make a lot of purchases but pay in full each month and escape interest charges.

THE ISSUE OF CUSTOMER PROFITABILITY

In an earlier chapter, it was noted that marketing is interested in only four goals: acquiring new customers, maintaining present customers, increasing the value of a customer, or migrating customers to other product lines. However, it is short-sighted merely to consider the income derived from a customer without also considering the cost of serving them.

To that end, the author suggests offsetting the margin contributed by a customer by the cost of acquiring and retaining them. The author balks at the actual calculation, suggesting it is a "complex financial analysis ... more sophisticated than the present discussion will allow."

(EN: it's not really that complex, but it is specific to the product, such that it would be difficult to discuss in general terms. I'd also proceed with a note of caution, in that one must differentiate between the costs that are caused by the customer versus discretionary outlays made by the company, and include costs that are germane only to a specific customer or segment.)

JIM.SHAMLIN.COM
DISCLAIMERS · CONTACT