Wilson Davis, Beverly Carey, and Joseph G. Roosevelt
PW Review, Fall, 1995
Segmentation is the marketing paradigm of the 1990s. Thanks to advances in data collection and analysis, organizations can accurately profile their customers and deliver a menu of products and services tailored to specific needs. However, that exercise, no matter how elegantly carried off, does not ensure profitability. Davis, Carey, and Roosevelt explore how activity-based cost analysis helps organizations determine which customer segments are profitable and which suggest that it may be time to “fire the customer.” Curiously, activity-based cost analysis is still underutilized, although it is a mature and powerfully revealing discipline. The model is from financial services; the implications are clear and urgent for many other industries.
When the tectonic plates beneath banking institutions started shifting about 15 years ago, senior management began to scramble for ways to maintain profitability. Fees for banking services, corporate and retail, began to soar. Excess expenses were cut without ceremony. Once these changes neared completion, product, division, and organizational reengineering began to take hold.
Many of these tactics did indeed enhance profitability. But in financial services those plates continue to shift. With prices up and costs way down, and with reengineering nearing completion, what will deliver the next generation of profits? The answer for many financial institutions has been to look more closely at their customer segments. Which of their customers are profitable and which are not? In this scenario, enhanced profitability is delivered by keeping customers that make money, and getting rid of—in effect, firing—those that do not.
To segment customers for profitability, financial institutions have had to understand costs. After all, how can you measure the profitability of a customer segment if you do not understand the costs generated by doing business with them? This need gave rise to a demand for activity-based costing, the accounting discipline that answers the question: What do all the activities cost that generate a product or service? The art and science of activity-based costing is not new; it is, however, enjoying renewed interest for the reasons mentioned above and because of the startling truths it is uncovering.
The notion of firing customers runs counter to the commonsense notion of what an organization is all about. And at the practical level issues often emerge which make the wholesale elimination of certain customer segments difficult at best. If the idea serves only to encourage consideration of the strategies required to transform unprofitable customers into profitable ones, it serves a good purpose. But we believe the idea has far more merit than that and should lead, in fact, to a practice of activity-based management.
While the next few sections of this article are developed within the context of a financial services organization, their discussion of implementing an activity-based cost system and responding to the strategic issues raised by it should offer insight regardless of discipline or industry.
The Way We Were
As a prelude to an actual case study, it might prove useful to explore some of the reasons why activity-based costing remains even today an innovative strategic weapon, not yet conventional practice. Hasnzzt fundamental understanding of costs always been in vogue? The answer is “not completely,” in the financial services arena. Intense regulation insulated many financial services companies, particularly banks, from the need to to be overly concerned about costs. In the final analysis, profits were regulated.
History, too, played a large role in shaping the orientation of accounting systems. Accounting as a discipline developed around industrial companies, and calculating the cost of widgets coming off the assembly line was straightforward: subtract out the cost of goods sold, and then allocate all other costs on a per unit or direct materials and labor basis. But the dominance of the cost of goods sold in determining profits had far-reaching implications. It meant that the purpose of financial and management accounting was oriented more toward valuing inventories, less toward allocating costs.
What about companies that didnzzt have inventory? For banks and financial services companies, Selling, General and Administrative Expenses (SG&A) were simply netted against the interest spread. The degree to which the spread exceeded SG&A was the chief concern of senior management—prior to the upheaval now occurring.
For the first 75 years of the twentieth century, domestic manufacturing firms—although sensitive to profitability—were somewhat insulated from the need to examine costs microscopically. Markets were right in their own backyards, and they were growing rapidly. However, when competition came from overseas firms that had cheap labor, no environmental regulation, and little if anything in the way of a social agenda in their cost structure, the interest in costs below the gross margin became particularly interesting. “Where precisely are all these resources being consumed?” became a matter of considerable curiosity among senior management of manufacturing firms.
Accounting systems were not designed to offer this kind of information, and activity-based costing, where conducted, was a special project rather than the instigator of a paradigm shift in the architecture of management information systems. Later in this article, we will return to the role of information systems and the potential of innovative system architectures to support activity-based costing (ABC).
A Case in Point
A banking client offers a case in point. In the late 1980s, management realized that the bankzzs check clearing service was losing more than $1 million a year. The client commissioned a strategic study to evaluate options for this business. Because the customer base was relatively small—approximately 11,000 customers—we could develop a compact model to explore the profitability of the customers, the products, and the organization. Rather than doing this on a unit basis, we emphasized the various processes under consideration. There were several different products or services in the check clearing business, such as clearing checks, issuing statements, and processing returned checks. Each product had several different elements which made up its entire cost, such as production, sales, distribution and administration. For present purposes, letzzs explore the core product, check clearing, and processes related to “producing” cleared checks. They were:
– Blocking and preparation of input
– Primary data capture
– Secondary passes
– Reject repair
– Fine sorts
– Federal reserve clearing charges
– Other clearing charges
In flow sequence, the production process is shown in Figure 1:
Figure 1. Sample Process Flow
To accomplish the translation from processes to a bill of material for each product or service, which would eventually reflect costs, we defined the frequencies in the production process. For example, when 100,000 checks were cashed, our client proofed 100,000 transactions and “repaired” 3,000 that were faulty for one reason or another. In total, utilization of the required capacities is shown in Figure 2.
Figure 2. Capacity Utilization with
Volume of 100,000 Checks
Process Process of
Resource Unit* Units/Expense†
Proof 1.00 100,000 10,000
Blocking 1.00 100,000 1,000,000
Primary data capture 1.00 100,000 200,000
Reject repair 0.03 3,000 5,000 1/$5,000
Secondary passes 1.20 120,000 200,000
Fine sorts 3.00 300,000 200,000
Transportation 1.00 100,000 1,000,000
Federal reserve charges 0.30 30,000 1,000,000 1/$6,000
Other clearing charges 0.00 0.00 0.00 0.00
* Reflects the capacity of the assets the company has on hand.
† Expense figure reflects the combined cost of resource units; i.e., for proofing, the 10 resource units cost a total of $1,000.
For manufacturers, this kind of modeling to determine gross profits is hardly new. But there is a new element here: extending to product and customer costs methods typically applied to logistics, distribution, service, marketing, financial administration, and information processing—the components of SG&A. Once this is done, a very different picture of profits begins to emerge than can be found by using direct labor or units as a basis for allocating costs.
When we had the process volumes in hand, we were almost home. The only remaining step was to build a model which would reflect the cost of cashing a check. This cost is shown in Figure 3.
Figure 3. Unit Costing Model
Check Bill Check
Expense Volumes Cost of
Proof $1,000 100,000 $0.01 1.00 $0.01
Blocking $2,000 100,000 $0.02 1.00 $0.02
Primary data capture $10,000 100,000 $0.10 1.00 $0.10
Reject repair $5,000 3,000 $1.67 0.03 $0.05
Secondary passes $10,000 120,000 $0.08 1.20 $0.10
Fine sorts $20,000 300,000 $0.07 3.00 $0.21
Transportation $2,000 100,000 $0.02 1.00 $0.02
Federal reserve charges $6,000 30,000 $0.20 0.30 $0.06
The full model was constructed by creating a matrix which reflected the costs of all the products and the processes required to produce them, including sales, distribution, and administration. The end result was painfully clear: The client was losing money on about 90 percent of its customers. When all relevant costs were allocated by activity, pricing was well below the cost of delivering products.
A unique facet of this case is that the bank had no preconceived notions about how to segment customers. In other words, the companyzzs services werenzzt offered as small business and corporate packages, or priced by volume or by the geographic location of customers. In our model, we were free to segment in any way that made sense. Figure 4 shows that segmenting generated a strikingly sharp-edged view of differences in profitability.
Figure 4. Profitability of Various Customer Segments
% of Current
Customer Set Volume Net Income
Current 100% ($1.2)
Outside State 11% $0.9
Within City Limits 17% $1.2
Smallest Customers 31% $0.3
Overseas Customers 23% $1.2
Largest Customers 23% $1.3
Correspondents 74% $0.1
Thus, our recommendation to maximize profits on the check clearing subsidiary was simply to do business with the largest customers. By doing so, the company would process just 23 percent of its current volume and increase profits to $1.3 million, a $2.3 million spread over current performance in the first year alone. We also recommended that the company not do business with the smallest customers, correspondent banks, customers outside the state, and overseas customers. In effect, the recommendation was to fire these customers.
At first, the client did not believe the findings. Management remained convinced that the company could still increase profit by increasing volume. And if by some chance we were right, there was the issue of relationships. The bank simply couldnzzt fire unprofitable customers with which it had a lending relationship. And then, how practical was it to turn loose 77 percent of that particular line of business? And what about employees? Banks just werenzzt downsizing the way they do today.
But time is the teacher. Our findings sat for several months. Then the bank accepted the facts and aggressively repriced services to unprofitable customers with whom it had no lending relationship. Customers driven to the door by the price increases were not detained: the books showed that their departure was increasing profits.
This case illustrates several important points about activity-based costing and its most promising consequence, activity-based management. The first lesson is a truism: knowledge is power. Segmentation is an important concept right now in banking and many other industries. Strategically significant segmentation requires knowledge of what it costs to do business with a segment. In this case, knowing how to segment customers was worth $2.3 million in the first year and $1.3 million every year thereafter—by any standard, a tremendous return on investment.
The second lesson is the benefit of statistical courage. It is not necessarily easy or pleasant to reprice customers to the point that they terminate the business relationship. But it is much easier when you know that these customers are taking their losses with them.
A third lesson is that the results of activity-based costing have to be viewed in the larger strategic framework. In this case, the conclusions needed tempering in light of lending relationships. The model stopped short of determining the profitability of the entire relationship the bank had with the customer.
This is not necessarily always so. Much depends on the formal scope of the activity-based costing project, on the organizationzzs information systems architecture, and not least on management intentions which have shaped that architecture. This bank, like many organizations, was vertically integrated, and the check clearing business maintained its own management information. To understand the total relationship between the bank and its customers, the analysis would have had to create process flows also for the credit side of the bank, and perhaps for the capital markets side as well. This could have been done. But the check clearing business had set the project scope and did not choose to invest in the more comprehensive analysis.
There was a fourth lesson—concerning time. This analysis and development of the model required 90 days from start to finish. Then the results sat in suspended animation for several months before the bank acted on them. It took another three months, conservatively estimated, to reprice services. And there then ensued another, less definable span of time in which customers thought about the new price structure and decided to take their (unprofitable) business elsewhere. Overall, it took up to two years to move from the activity-based cost analysis to the restoration of profitability. This may seem extreme; unfortunately, it is not altogether rare.
Systems Architecture and Activity-Based Costing
The scope and efficiency of activity-based cost analysis depend in part on an organizationzzs information systems architecture. For those who recognize the strategic potential of activity-based costing, a sequence of questions looms large: “Are our systems adequate to the task? What changes are required? At what cost?”
It is worth exploring the magnitude of change required to embed activity-based costing in the infrastructure of a business. In many organizations, profitability measurement systems are general ledger-based, similar to the one depicted in Figure 5.
Figure 5. General Ledger-Based Profitability Measurement
While general ledger-based systems have the advantage of size and processing power, there are limitations. First, they may not be even remotely capable of handling product profitability on an activity basis. And general ledger systems absolutely cannot process the volumes of data required to determine customer profitability, at least in a large commercial bank. Finally, extracts of the information required for ABC analysis are often difficult to achieve.
Are PC-based profitability measurement systems better adapted to the needs of ABC? Consider Figure 6.
Figure 6. PC-Based Profitability Measurement Architecture
For all the advantages of the end-user interface, this architecture also brings difficulties. There is a continual need to balance and reconcile to the general ledger. And then, the task of integrating the PC system with the budget system for actual versus budget reporting is arduous. Finally, a single PC for profitability measurement is often at odds with the LAN configuration of most finance departments.
New information technologies are emerging which allow for a radical change in system architecture. In newer architectures, a data warehouse exists on equal footing with the general ledger. While the general ledger remains unencumbered for legal reporting purposes, the data warehouse encourages analysis at other functional levels for purposes of decision making. Activity-based profitability measurement is just one by-product of the data warehouse shown in Figure 7.
Fig. 7. Data Warehouse-Based Information System Architecture
Much of the data required for senior-level decision making and analysis is actually trapped inside the legacy or mainframe-based information systems of many organizations. General-ledger driven systems tend to summarize data according to a chart of accounts, thereby obscuring the data needed. By contrast, the data warehouse creates consistently defined, highly discrete units of data and “warehouses” them in a single, accessible place to be pieced together not as legal reporting dictates but as the analysis at hand requires. This extraordinary versatility is a major change in an organizationzzs system architecture—but achieving it is costly in dollars and hours. Therefore, zeal must be tempered by caution.
Are the risks and costs worth the potential benefits? The answer, unequivocally, is yes. In financial services, organizations are beginning to realize they no longer have the luxury of time to create and maintain profitable customer segments. In the two years it might take an organization to tinker its way to profitability, a faster-paced company will reshape whole markets through targeted, innovative, and astutely priced offerings.
Activity-based analysis and the resulting segmentation make an organization more efficient and profitable. They provide the basis for the skill and perspective we call activity-based management. What are some of the key precepts of this approach?
Repackaging. Many times organizations will find that they can achieve profitability by unbundling their products. To see how this can work, consider checking accounts in terms of customer use of teller services versus ATMs. Checking products have typically been priced identically, or nearly so, for all customer segments. However, activity-based analysis is likely to uncover that one customer segment never sees the teller, another uses teller services once each week, while still another makes extensive use of teller services.
What is happening here? The segment that makes extensive use of teller services is in effect subsidized by the segment that never does so. The initial design of the product rarely envisions this behavior. Rather than continuing to allow one product to address all of these behaviors, activity-based management suggests unbundling the product and letting customers pay on an à la carte basis. Customers who rely on teller services would pay fees commensurate with the resources they consume. Those who use ATMs also pay fees reflecting the resources put at their disposal.
Activity-based management thinking suggests a further step. Suppose that an à la carte menu of services enhances profitability for both segments, but the ATM segment is demonstrated to be three times more profitable than the branch customer segment. What should management do? Provide incentives to get the branch customers to become ATM customers. In this instance the bank is not abandoning or firing branch customers, nor is it any longer subsidizing one population through another. The bank is simply maximizing profit. Now thatzzs activity-based management!
Limit Options. Customer segments that buy standard products are almost always more profitable than customer segments that demand options. When the costs of delivering options—that is, red versus standard black widgets—are allocated on a unit basis, the impact on profitability may not be obvious. But when costs are assigned by the activities required to produce the option, then an altogether different picture emerges. The activity-based analysis will show that while gross profits are comparable, operating profits are considerably thinner if not negative: optional red widgets chew up notably more resources than standard black ones in administration, purchasing, logistics, distribution, and after-sales service.
Automobile manufacturers appear to have adjusted product strategy in accord with this insight. Twenty-five years ago, options abounded. Each model came with a V-6 or V-8, hardtop or convertible, air or no air, automatic or manual transmission. Today, the auto companies seem to have reached a compromise with their customers: there are fewer options, but the standard model has air conditioning, cassette stereo, airbags, automatic transmission, and anti-lock brakes. And Big Three automaker profitability has soared.
Some companies are masters at managing customer options. At fast food giant McDonaldzzs, customers can show up at any time of day and order any combination of the 25-plus items on its menu. McDonaldzzs dampens the effects of this behavior by bundling typically preferred options into its famous “Value Meals.” Customers get a break on price, and that vectors them toward the standard combinations. McDonaldzzs gets a huge reduction in production costs by diminishing the number of options demanded and by creating a large customer segment that purchases standard Value Meals.
Even in banking the same principles hold. One banking client found that the segment of customers who use the bank on Saturdays is its most profitable. By contrast, the customer segment that uses the bank after hours is least profitable. In response, the bank increased hours on Saturdays and decreased after-hours availability.
Improve the Quality of Service. Profitability sometimes languishes because products remain undifferentiated in the customerzzs mind. Differentiating the product and raising prices can make profits leap. For example, at many financial institutions the same checking account product services low and high net worth customers. Not surprisingly, low net worth customers donzzt really appreciate the product while high net worth customers feel underserviced. If the activity-based costing model shows that both segments are profitable, managementzzs options are reasonably clear. Increase the quality of the checking account services for the high net worth segment so that the product is differentiated in their minds, and charge accordingly to maintain or increase profitability. On the other hand, trim some features for the low income segment and reduce their fees. After all, why give customers a Mercedes when they are in the market for a Honda Civic, or vice versa?
Developing New Products. A close corollary of upgrading products is the development of new ones. Once an organization has defined a profitable customer segment, the risk of developing new products and going after new customers in that well-understood niche is dramatically reduced. At some financial institutions, when a high net worth product is developed—consisting perhaps of checking, a credit card, wire transfer facilities, brokerage services, and account reconciliation for tax purposes—it is sold wholesale to affinity groups such as partners in professional service firms.
New sources of profitability are not restricted to pricey products and the high net worth segment. In the retail world, for example, customer segments that tend to return products under warranty may be the best candidates for extended warranty programs. In the fast-moving world of software, customer segments that tend to call for product support will probably welcome prepaid support plans. In banking, the segment of customers who frequently bounce checks would at first blush appear to be candidates for firing. But they are also the most likely segment to purchase overdraft protection—and when they do so, the segment may no longer be unprofitable.
It is not just the pricing side of the equation that delivers profits. Activity-based analyses will identify excessive costs and process redundancies. If you are overstaffed, that should become evident, just as the need to “fire customers” or convert the relationships became evident. Whichever side of the equation you scrutinize, the stakes are high. All the more reason to have the most accurate possible measures of costs and profits.