Executives with Fauquier Bankshares in Warrenton, Va., know all too well that relatively few of their customers account for most of the bank's profits.
And they believe that customer-specific profitability data can help their effort to retain the best customers and cut their losses on others.
But the bank's path to profitability measurement has been a twisted one, marked by a host of tough decisions.
The bank has considered software systems with distinctly different approaches. And it continues to question the usefulness of precise customer profitability data, given the bank's long-standing aversion to aggressively marketing its services.
Community banks across the country, experiencing similar complications, are learning that turning customer profitability figures into concrete marketing action requires far more than picking the right technology.
"Few if any banks under $1 billion of assets have true customer profitability information or the culture to enable them to use it effectively," says David Koehler, chief financial officer of Fauquier Bankshares.
Still, bankers and industry observers say using profitability information to rethink product pricing and design, to target marketing campaigns more effectively, and to restructure compensation programs can bolster organizational profits and help fend off nonbank competitors.
Those are among Fauquier Bankshares' goals. The bank got started and then abandoned one system to measure the contribution of individual customers. But it has begun to take some of the steps commonly associated with customer profitability while it works toward installing a different system.
Its first is to make telephone calls to what it believes are its best customers.
"We will initiate a dialogue to at least make sure they know we're here to serve them," Mr. Koehler says.
The bank is implementing performance-based compensation. It will encourage customers who bank with numerous institutions to consolidate their business at Fauquier by offering free checking in exchange for a minimum balance of $2,500 in a combination of deposit accounts.
When a glitch with a new debit card processor resulted in many customers' cards being rejected in January, the bank's executives thought they had better compensate customers for their inconvenience before complaints rolled in. But rather than extend the courtesy to all card holders, the bank targeted those it believes are most valuable.
Other banks are targeting high-transaction, low-balance customers for fee increases or repricing loans and deposits based on a fuller understanding of the cost of supporting the product.
First National Bank in Goodland, Kan., for example, decided it offered municipalities too good a deal on time deposits after factoring in overhead costs, says Lawrence McCants, president of the $260 million-asset bank.
It has cut rates on municipal deposits by more than 1% on average since installing a profitability measurement system sold by Financial Technology Inc. of Chicago about three years ago - and generally without losing transaction account business.
Fauquier is well on its way to more sophisticated decision-making, says Mr. Koehler. But it has made a few extra turns so far. The bank bought and then junked a marketing central information file system - or MCIF - developed by Harland Financial Marketing Services of Orlando.
It began installing a series of profitability measurement tools built by IPS-Sendero of Norcross, Ga., but then backed off and focused instead on learning to sell.
In these steps alone, the $220 million-asset bank grappled with several issues critical to buying technology.
For example, banks often choose whether to use an MCIF system, a customer profitability system or a combination of the two. There are clear trade-offs.
At the heart of MCIF systems - Harland is a leading vendor - is a mechanism to match members of a household when names and addresses are not identical.
That is crucial for the bank that does not want to alienate the holder of a profitable account if another household member has an unprofitable one. MCIFs, which come loaded with many key financial assumptions, may be especially attractive for small banks with small accounting staffs.
However, MCIFs use methodology to calculate the profit of customers and households that may not be entirely accurate. For example, they typically allow, or even encourage, a bank to subtract industry average costs supplied by the Federal Reserve System from customer-generated revenues.
Many observers find that practice troubling.
"Sometimes bad information is worse than no information," says Peter G. Humphrey, president of Financial Institutions Inc., a $1 billion-asset bank company in western New York now shopping for a measurement system.
Customer profitability systems - Financial Treasury and IPS Sendero are among the leading vendors - are more sophisticated than MCIFs.
But they are not necessarily configured to identify crucial household relationships. In addition, they require banks to identify their actual cost of completing transactions or maintaining accounts. That means allocating overhead expenses to lines of business.
Trouble is, as Fauquier executives discovered, allocating overhead expenses can provoke turf wars before mid-level managers even buy in conceptually to the whole exercise.
Expense allocation is tricky, says Jerry Weiner, chairman and chief executive of IPS Sendero, who argues that producing the average cost of performing a particular function is not nearly as useful as the incremental cost.
Still, the issue is not necessarily the precision of information provided by software.
"Calculating profitability accurately is only one piece of it," says Mary Knox, an analyst with Gartner Group/Mentis Financial Services in Durham, N.C. "The other pieces are applying it appropriately and interpreting it appropriately."
Bankers should be wary of technology vendors who suggest that calculating profitability is straightforward and that they need do little more than plug numbers into a model, Ms. Knox says.
Even with highly detailed quantitative information, banks need better qualitative information about what motivates customers to act as they do, says IPS Sendero's Mr. Weiner.
Recognizing an opportunity to begin segmenting customers, Fauquier has identified interim steps on the road to completing installation of the IPS- Sendero system. The bank is introducing performance-based compensation, which it regards as fundamental to creating a culture of selling. And it is basing some of its early sales efforts on broad inferences about profitability.
Compensation will be tied more closely to performance as the bank improves its ability to measure customer profitability.
Although measuring profitability is complicated, it is safe to say that high balances generally correlate with greater profits, says Mr. Koehler, the chief financial officer.
Fauquier is using a data warehouse that it bought from M&I Data Services of Milwaukee to aggregate balances by household.
Then it is assigning officers to call the highest-balance customers, not to pitch specific products but to nurture a relationship.
In the long run, compensation should not be based on the number of accounts an officer opens or dollar volume lent out, Mr. Koehler says.
But that is where Fauquier will start, to get employees used to receiving pay for performance. Ultimately, pay should be based on the profitability of business that bankers generate or retain, he says, adding that Fauquier would build the ability to measure customer profitability at a pace that allows staff members to grow comfortable with judgments underlying the measurement system.
After a bank masters measuring organization and product profitability, Mr. Koehler says, it must develop a funds transfer pricing mechanism, which gives credit for the interest revenue ultimately supported by a deposit account or the interest cost borne indirectly by a loan account. As with every other step in the profitability computation, bankers need to beware of vendors who assure them that funds transfer pricing is a science.
Mr. Koehler says MCIF systems often will measure the profitability of a loan account using an average cost of funds at the time a report is generated. He insists that only the cost of funds that supported a particular loan when the loan was written is valid.