Kim Davis, Executive Vice President ofAIMS, sat in her 43rdfloor corner office overlooking the Manhattan skyline, reflecting on the challenges facing the investment services business in 2000. Profits had come easily during the longest economic expansion ofthe century. However, signs ofweakness in the economy, financial market volatility, intense competition for high net worth customers, and the proliferation of complex technology-dependent products were all making her life much more complicated AIMS had recently invested in new analytic tools to help think more strategically about its operations and customers. Kim wondered how much the new analytic approach would really impact business decision-making. Was intensive customer segment analysis a real opportunity or just another "shot in the dark?" AIMS is one of the larger investment services providers in the U.S., approaching $500 billion in assets in 2000. Of this total, a little more than half was in mutual funds and the balance in brokerage accounts. This case deals with customer profitability assessment for AIMS' 3.9 million households, up from 1.8 million in just four years. Until I 999, AIMS had no system for measuring the profitability of any specific customer. SEGMENTATION AIMS spanned two separate and very different product lines (mutual funds and full-line brokerage services), but that was only one element of the complexity it faced. In addition to this product complexity, it also spanned three distinct "distribution channels" (Call Centers, Full Service Branches, and E-business), and a complex array of customers with diverse asset holdings, trading patterns, investment objectives and service requirements. There was no particularly sharp focus on what kind of households to add. The basic idea was high wealth, but that was not pushed exclusively at all. Basically, AIMS wanted to do business with the same 2 million American households (over $1 million in invested assets) that 2 I other major financial services firms were pursuing. In I 999, AIMS introduced segment analysis, starting with a four-way segmentation that mixed three different dimensions: asset holdings, trading activity and age (as a proxy for investment objectives). The first segment was any household with more than $500,000 in assets under management at AIMS ("High Net Worth," or "HNW''). Failing this test, the second segment was households trading more than 36 times in I 998 ("Active Traders," or "AT"). Failing this test as well, the third segment was households where the principal customer was already retirement age (60 years old). Finally, customers failing all three of these tests comprised the fourth segment-all other, termed "Core" customers. "Core," with more than 70% of all households, was the largest segment. The primary role of any segmentation is to facilitate analysis leading to management actions tailored to the specific needs of defined customer subgroups. No particular segmentation is ever beyond dispute. Whatever approach is chosen necessarily emphasizes some distinctions and deemphasizes others. But, the AIMS segmentation was particularly contentious on two grounds: I) it segmented current customers rather than a market. It is as if Procter & Gamble were to segment the detergent market based on how many pounds of Tide are purchased; 2) the sequence specific classification scheme meant that labels could be misleading: for example, the segment Active Trader applies only to households which are not each HNW. And, "Retiree" applied only to households which were not each HNW or AT. 2 AIMS FINANCIAL RESULTS As shown in Exhibit I, AIMS did quite well in 1999. Net margin afteJ tax was about $156 million on an underlying equity investment of about $625 million. But, 1999 represented the height of the prolonged bull market. The year 2000 was projected to be much less bullish, and most Wall Street observers envisioned the next few years to be much less rosy than the previous ten. Even in 1999, performance was not consistent across all the customer segments. Pre-tax margin ranged from a high of48% for HNW, to only 6% for Retirees and minus 4% for Core. The revenue breakdown across segments in Exhibit I is based on actual identification with individual customers. The expense breakdown starts with an annual "unit cost" study that uses "Activity-Based Costing" (ABC) principles. The study first assigned all operating costs from the General Ledger to specific processes or "activities." Then, the activity costs were divided by throughput measures for each activity, to create "cost per unit of activity" for each sub-stage of each process. This process is illustrated in Exhibit 2 for estimated costs for 2000. Individual unit costs were then multiplied by throughput totals for each segment and aggregated to provide total expenses per segment as shown in Exhibit I, a report format which was new at AIMS in 1999. THE, CUSTOMER/PRODUCT PROFITABILITY INITIATIVE As a management report, Exhibit I was too aggregated to identify actionable issues. In 2000, AIMS undertook a project to take customer/product profitability reporting down to the individual household level to provide more useable, timely, and integrated information for decisionmaking. The new system combined unit costs from the annual ABC study with current actual household activity and attributes (e.g., products held, services used, number of trades, number of rep-assisted phone calls) extracted from the Marketing Database to generate profitability by household. The data then were exported into easily queried online analytical processing (OLAP) "cubes." OLAP cubes allow profitability analysis of the intersections among customer attributes, product/service attributes, and channels ofdistribution. Exhibit 2, which illustrates the first step in this n.ew .system (unit costs across processes), is highly s1mpltfied for purposes of the case. As shown, a "driver" was chosen to proxy the activity in each processtelephone calls as the driver of activity in the Call Center, for example. Next, a count was made of the total estimated units of the activity for 2000 for each driver7~.l million calls for the Call Center, for example. Fmally, the total cost for the process was divided by the total activity count to calculate cost per unit of activity for that process. Some of the assignments ofcosts to activities and some of the activity measures are "soft," but the activity costs tagged to individual households based on actual household activity are conceptually plausible and at least directionally correct. Similarly, product-specific and service-specific revenues are driven down to a household level. Household profitability calculations are thus based on actual asset holdings, fee-based services consumed and activity usage. The actual system in use allows for 11 categories of customer revenue and 70 categories of process cost. Conceptually, Exhibit 2 represents "long-run average cost" for each activity. It does not attempt to portray marginal or incremental cost because it is not intended for use in short-run cost-volume-profit (CVP) analyses. Since very little cost at AIMS is variable with short-run volume fluctuations anyway, short-run CVP analysis is really just based on revenue changes. Almost all costs are "step costs" which go up (or down), in chunks as capacity is added to (or deleted from) the system. In a business as fast-growing as AIMS has been in recent years, capacity is typically being added every year in many places across the process value chain ahead of usage requirements. Thus, there is almost always excess capacity in the system. And, the extent of excess capacity varies across processes, depending on where growth has been fastest and where recent expansions have been made. The analysis in Exhibit 2 divides current cost by current throughput to calculate unit cost. The analysis thus charges any excess capacity to the current users of the process. This is debatable, conceptually, but is not recognized as a practical problem at AIMS. The expense base grew substantially faster than throughput volume between 1995 and 1999 in anticipation of even greater future growth. In 1995, ~here was about 10% excess capacity (on average) in the operating expense base. Capacity grew at a compound rate of about 26% from 1995 to 1999, versus households growth at about 21 %. As a result, excess capacity in 1999 was a much larger percentage of the expense base, across branches, the call center, on-line activity, transactions processing and account maintenance activity. Kim wondered how much of operating capacity was devoted to unprofitable customers. THE SEGMENTATION REFINEMENT INITIATIVE Another new initiative in 2000 to enhance customer profitability analysis involved further refining the segmentation. The goal was to better identify customer clusters that would be responsive to specific managerial actions. Kim Davis was chairing the task force coordinati~g this effort. The primary four-way segmentation was expanded to 11 categories as shown below. 3 AIMS High Net Worth(> $500,000 ofassets under management) I. (16,000 Households)> $2,000,000 in assets under management 2. (141,000 Households) -$500,000 to $2,000,000 Active Traders(> 36 trades per year) 1. ( 9,000 Households) -more than 200 trades 2. (12,000 Households) -60 to 200 trades 3. (19,000 Households)-36 to 60 trades Retirees 1. (262,000 Households)-$100,000 to $500,000 in assets under management 2. (607,000 Households)< $100,000 Core 1. (426,000 Households)-$100,000 to $500,000 in assets under management 2. (1,762,000 Households) -"Boomers" (40 to 59 years of age) 3. (434,000 Households)-"Young Professionals" (under 40 years of age) 4. (192,000 Households) -All Other, including employees CUSTOMER PROFITABILITY ANALYSIS As noted earlier, although the company was very profitable in 1999 as the ten-year bull market continued, the senior management group was concerned about the tremendous range of profitability across customer segments and about the potential for substantial profit erosion when overall markets slowed down, as was widely anticipated over the next few years. Kim challenged the management team to analyze customer mix carefully to identify problem areas and potential corrective actions. One new management report now being produced each quarter showed income statements for each of he eleven segments broken down by deciles, starting with the most profitable 10% of households and ending with the least profitable 10%. Not surprisingly, the tenth decile in all eleven segments was unprofitable, even before considering any allocation of marketing expenses directed at acquisition of new customers. It was generally agreed that profitability analysis of current households should exclude all expenditures directly related to new households-either "prospecting" expenses in marketing or new account set-up expenses in the back office. When the segmentation was ignored, 75% of the bottom decile customers were in the Core segment and 80% had less than $100,000 in assets under management. The wide range ofprofitability across deciles and segments is summarized in Exhibit 3 for 1999. The aggregate loss on all unprofitable households in 1999 was $248 million. Obviously, unprofitable households are an important concern for AIMS. Kim Davis wanted to identify the roots ofthe problem as clearly as possible. At a casual level of analysis, an unprofitable household suggests one oftwo responses: "Fire" them, because AIMS does not want customers on whom it loses money.

Questions:

1. What is your opinion of the process the firm followed to assess the customer profitability?

2. How would you use this information to improve your business?

3. Which are the potential sources of hidden profitability that you detect for AIMS?

4. Why is the firm facing this situation? How can this be prevented towards the future?