If you fork over the extra money to fly first-class, you deserve to be pampered. Through its new segmentation strategy, Banc One Corp. is applying that concept to its retail customer base.

Soon, the $87.8 billion-asset bank will complete defining its six million households according to the degree of profitability of each. It also will segment them by banking behavior preferences.

The aim is to examine customers in order to understand not only how profitable they are but also how they got there.

And since it now knows who makes the most money for it, the company can roll out the red carpet to make sure those customers stay.

Most banks have failed miserably in efforts to foster loyalty among their best customers, according to Fred Reichheld, a director at Bain & Co., a Boston-based management consulting firm. The reason: "Banks stand out as one of the weakest industries in effectively segmenting their customers."

In contrast, he cited Staples, the office supply superstore. It has refined its customer tracking system to the point where it knows thoroughly the top 5% to 10% of its customers. "As a result, it understands - throughout a life cycle - what the best customers buy, where they buy, and why they buy," said Mr. Reichheld.

With that knowledge, the company can give better service to this group. "It's making money with these insights," Mr. Reichheld added.

Banc One is aware of the danger posed by not getting to know its most profitable customers. "One of our biggest fears is that we stuck a lot of people way back in coach when they have been paying for first-class," said Craig J. Kelly, senior vice president and director of marketing.

Like many bank companies, Banc One had already segmented its retail customers on the basis of demographic characteristics such as age and income level. These data, though, are useless for determining customers' profit contribution, according to Mr. Kelly. Customers who are quite different demographically can end up in the same profitability segment, he said.

The paths of diverse customers to the same profitability segment can vary, Mr. Kelly added. For example, checking account customers deemed profitable may carry high balances or pay high fees.

The bank has fine-tuned its profitability measurement almost to the penny. In the process, it has discovered that the old 80/20 rule no longer applies.

"We used to think that 80% of our profits came from 20% of our households," Mr. Kelly said. "We suggest it is far worse than that. In fact, our higher-profit households may in fact represent in excess of 100% of our profits because the unprofitable ones subtract so much."

Profitability of the bank's households is calculated on a fully loaded basis, as well as on an incremental basis. Under the fully loaded method, all overhead costs, like employee salaries, are allocated.

Since these costs are ultimately borne by the customer, understanding profitability in this way is useful for pricing products and services. But since fully loaded costs remain the same regardless of whether a customer stays or leaves, this approach is less useful in formulating retention strategies.

The incremental cost approach factors in costs incurred in serving customers. These include mailing monthly checking account statements and offering service at a branch, automated teller machine, or via phone. With these calculations, the bank can determine, among other things, what costs would disappear if certain customers left.

Software developed jointly with First Manhattan Consulting Group helps the bank collect and churn information in a data warehouse that runs an Oracle relational data base.

Information includes how, where, and how often customers conduct banking transactions. In addition, data from the bank's various mainframe systems, along with information from a marketing customer information file, or MCIF, are fed into the warehouse.

The software then clusters, or segments, the bank's households according to levels of profitability indicated by numerical codes. The number of household segments currently stands at 72. Views of profitability at the account and customer levels are also provided.

Now that the bank knows more about customers, it is devising ways to target profitable customers with better service and products.

For example, it may consider dropping blanket advertising of certificates of deposit and their rates to the general public. Instead, CDs would be priced based on a customer's profitability ranking.

"We can say, based on the relationship you have with us, 'This is what we can afford to pay you,' " Mr. Kelly explained. The result: many markets of one versus one market of many.

The segmentation program also will help the bank increase cross-selling to its profitable customer base. It's on the right track here, according to John DeMarco, senior vice president at Payment Systems Inc., Tampa.

"For too long, banks have been coming up with excuses as to why they can't expand their relationships with their best customers," Mr. DeMarco said.

One of those excuses is that banks often mistakenly believe they have all the business they can get from this customer base. And they are reluctant to bother high-end customers with sales calls, said Mr. DeMarco.

Banc One's strategy calls for it to be more selective when deciding to whom it will cross-sell - thus abandoning the industry's age-old strategy of blindly increasing cross-sales ratios regardless of the financial return that may result.

"We have found that the more you cross-sell certain households, the more unprofitable they become," Mr. Kelly explained. "We want to prevent the danger of cross-selling our way into a deeper hole."

Mr. Kelly emphasized that while the bank will focus its cross-selling efforts on more profitable households other customer segments won't be denied products if they request them.

Nor is it the bank's intent to impose higher prices on less profitable customers to drive them out.

Rather, the profitability-related information will help the bank attack weaknesses in its own cost structure and in its internal operations. Indeed, the bank is currently reorganizing its highly decentralized retail operations to boost efficiency and cut costs.

"We've made mistakes in how we've priced some of our products," Mr. Kelly acknowledged. "Should the customer be penalized because we goofed? Absolutely not."

Although the bank does not intend to shed its less profitable customers, it may consider changing product mixes for these households, changing product and service pricing, or coming up with alternative product offerings.

Banc One will be able to determine what those changes should be since it has a better handle on how its customers do their banking, according to Bobby Mehta, a vice president at Boston Consulting Group.

"If you understand how customers behave, and how they prefer to interact with you, you can redesign core product offerings" to offer the customer incentives through pricing, Mr. Mehta said.

He proposed an example: a 20-something customer working at a first post- college job who does not make a lot of money, generates a lot of transactions, does not visit branches often, and always revolves credit card balances.

Since the bank won't be able to break even on just the customer's checking account, it should target this customer type in two ways, Mr. Mehta said. First, get his or her credit card business. Then offer a checking account that features low or no fees for not coming into the branch.

By targeting only those customers that fit the above behavior profile with such a pricing strategy, banks can avoid alienating the customers with dissimilar profiles, Mr. Mehta explained.

Obviously, like most bank companies, Banc One will continue to get traffic in its branches. There, it will exploit its segmentation scheme to improve customer service and refine cross-selling efforts. The bank is now programming its platform systems to instruct customer service representatives automatically on appropriate actions for the segment to which a customer has been assigned.

The 72 segments identified by the system will be consolidated into four broader segments at the branch level to prevent information overload.

The bank, Mr. Kelly explained, wants to make sure that employees know which are the bank's most profitable households so that they will get superior service. Often, he added, customers believed by bank employees to be the cream turn out to be anything but.

In addition to getting instructions on how to serve profitable customers in the branch, employees are being given lists of these elite customers.

"We're telling staff, 'You'd better get to know these customers, even though many of them don't come into the bank,' " Mr. Kelly said. Employees are being told to reach out and touch this group by, for example, phoning to thank them for their business.

Some at the bank work with all 72 household segments, doing sophisticated data mining and modeling for applications such as direct marketing strategies. Another group - marketing and sales strategists - work with 20 segments consolidated from the 72.

The segmentation system will eventually be linked to the bank's sales management program. Once that's achieved, sales people will be able to decide which customers to call and which products to pitch to them.

Currently, Banc One's data warehouse runs on an Intel-based file server. The company is evaluating whether that platform will continue to be adequate.

The bank is also working on smoothing the currently cumbersome configuration of its systems. For example, employees now must pull data from the mainframe and transfer them to the data warehouse. Moreover, the MCIF is not fully integrated within the warehouse. Soon, mainframe data will be pulled automatically, and the MCIF will become an integral part of the data warehouse.

Also underway is expanded use of modeling software, which helps the bank find customers likely to respond positively to marketing campaigns. Even before the segmentation scheme was adopted, the bank's direct marketing group had succeeded with this technique by reducing the size of direct mailings while boosting response rates.

Now that the segmentation information is available, Banc One will be able to formulate models that will deliver more precisely targeted direct mail campaigns for even better returns. In addition, the bank is now able to track marketing campaign results by segments.

As Mr. Kelly noted, "The more ways we can cut the information, and isolate who responds, the better the next direct mailing will be."

Banc One also wants to see whether customers move from one profitability segment to another over time, and why. Such unanswered questions are part of the reason the company does not want to start pricing customers up and out.

"It could turn out that ultimately we were doing that to the wrong households," said Mr. Kelly.

Other techniques the bank is exploring include neural networks for use in direct marketing. The bank, though, still hasn't exploited all the information that traditional statistical analyses make available.

"There's so much information involved," said Mr. Kelly. "First, you have to go after the low-hanging fruit that's easy to reach using traditional methods. Once you've grabbed that, you try to reach the top of the tree."

Ms. O'Heney is a freelance writer based in New York.

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.