How Should We Measure Customer Profitability?

David McNab, President, Exchange Synergism Ltd.

In the previous article in this series we discussed what customer profitability is and some of the key benefits of measuring it. We argued that you can ill afford to manage without this information as you make decisions concerning the sales, service, product management, operations and finance functions of your business.

In this article we bring the discussion down to a more concrete level, exploring the measurement issues you are likely to encounter in the design of a customer profitability measurement "formula." We talk about these issues in the context of business decisions, because the appropriate measurement method depends on what question you are trying to answer. Unlike public financial reporting, where accounting rules are prescribed, customer profitability measurement is management information. There are no set rules that have to be followed. The choice of accounting methods depends on the information needs of internal decision makers.

To build a model that will meet your needs, you need to understand the relationship between the types of decisions you make and the accounting choices that have to be made when building a measurement model. If you don't consider these factors you run the risk of (a) building an expensive model that doesn't help you run the business or even worse (b) making misinformed business decisions that have a direct impact on your customers.

Types of decisions

Customer profitability can be useful to inform decision making in many contexts. Regardless of where responsibility lies within your organization, decisions are made routinely in which the value exchange between customer and company is being changed. Measuring customer profitability gives you insight into the effects of these changes and offers you the opportunity to manage them.

Some of the proactive decisions that marketing managers face every day deal with the allocation of resources, specifically things like:

  • Choosing which prospective customers to acquire
  • Choosing which customers you want to retain
  • Choosing which customers to cross-sell (and what to sell them)
  • Choosing which customers you may want to abandon
  • Choosing channels of distribution
  • Setting the price (or discount structure) for a product or service
  • Setting sales compensation rates
  • Setting reward programme entitlements
  • Setting advertising and promotion budgets
  • Setting service levels for customers
  • Identifying customer behaviours that create or destroy value

In addition to making these proactive choices, we also want to know what kind of results we are producing. Customer profitability can provide us insight into key performance metrics like:

  • Measuring the cost of lost business
  • Measuring the value of new business
  • Return on a campaign
  • Return on an offer
  • Return on a pricing change
  • Return on changes in service delivery

Bases of measurement l: time dimension

There are four primary measurements of customer value that can be used to help us answer these questions.

Historical value of a customer looks at the value earned from a customer relationship over an extended period of time, such as prior fiscal quarter, prior year or since the start of the relationship. It can be measured as a simple average of previous periods or can be time weighted, placing higher emphasis on recent periods. Averaging in this manner has the effect of smoothing reported results for a customer, lending consistency to the reported values. Historical value is most useful for comparing customers with each other in order to rank them, selecting groups for marketing efforts and for assessing pricing or budgeting decisions.

Current value looks to a shorter time frame, often a month (in order to coincide with reporting cycles). Current value is often volatile, since cyclical factors in the relationship are often not reflected within a single month. Current value has the advantage of highlighting the effects of changes in the customer relationship when compared to previous period current values. It is most useful for quantifying the benefit of campaigns, new offers, and pricing changes on customer value.

Present value is a future oriented measurement, which typically considers the future revenue and cost streams of the customer's existing business. This measure is usually only extended to include the contractual lifetime of ongoing products or services. Revenues and costs are projected into the future using essentially the same methods as in current value and then are discounted to the present using a discount rate appropriate to the business (for example the cost of capital). Present value is useful for ranking customers according to value, determining sales compensation rates, and is frequently used as a basis for modeling the impact of decisions concerning price and service before they are implemented. A comparison of present value of customers between time periods also provides a sound basis for evaluating program results for marketing campaigns.

Lifetime value is another future oriented measurement, again based on the same methods that are used in current value measurement. What distinguishes it from present value is a modeling component: lifetime value takes into account projected revenue and cost streams not only from the existing relationship but also from business that is expected to be done with the customer in the future. To implement lifetime value a company must have a substantial insight into renewal or repurchase behaviour and propensity of the customer to add or reduce the extent of their business relationship in the future. The projected cash flows are usually discounted in a manner similar to present value. Lifetime value is often viewed as the optimal customer value measurement for most decision making situations. Unfortunately few companies have the experience and models required to produce a lifetime value measurement that management is comfortable with.

There are different measurements for different purposes, and variations in precision levels that can be attained. In an unlimited resources situation, we would provide management with all of them. Best in class customer profitability systems today support all of these measures. Implementation is primarily limited by the available data, models and resources of the company that is implementing the technology solution.

Starting out, most companies strive to implement a credible current value measurement. Under ideal circumstances this can be extended backwards in time to encompass historical information, generating multi-period results in the first release of the system (six or twelve months of history is usually the best that can be done practically).

From this base of current value history the next step is usually implementation of historical value. Present value and lifetime value are more sophisticated measurements and are usually implemented as a later upgrade to customer profitability information after the organization has gained experience working with the new information provided by current and historical measurements.

Bases of measurement II: accounting theories

Regardless of which bases of measurement are selected, there are a number of technical accounting issues which will need to be addressed when tailoring the calculations to fit your business. In this section we will explore some of the major decisions that need to be made, looking at revenue, cost, and finally probabilistic items like insurance claims, coupon redemptions and loan losses.

As a Marketing professional you may not view accounting as an area of personal interest or expertise, and you may be inclined to delegate determination of accounting treatments to the Finance department. This is not always a good idea. The way customer profitability is calculated may need to be different than accounting profitability measurement in order to be useful for marketing purposes. For example, consider account closing costs; from a financial accounting perspective, they cannot be included until the account closes. From a marketing perspective, however, we need to include these costs during the lifetime of the account when we are setting prices. This type of conflict needs careful resolution when designing how your customer profitability calculations will work.

Another consideration is the degree of precision that you need to have in order for the measurement to be useful for marketing purposes. If you are sponsoring development of the measurement you certainly have a voice concerning how far your company should go to make the measurement reflect all revenue and cost items fully. In some industries it may be appropriate to simply measure revenue or revenue minus some simple costs (such as cost of goods sold) to provide information which adequately identifies the relative profitability of customers. In this article we contemplate a more sophisticated - and complex - model in order to make the discussion more universal. Your knowledge of your business will ultimately determine how much detail and precision needs to be included in your own customer profitability measurement.

If you want to have a useful customer profitability system, you will probably need to work with Finance as your partner to provide expertise defining the requirements that will be implemented and to help resolve these types of conflicts.

Revenue

The main choice that needs to be made concerning revenue is whether you want to recognize it on a Cash basis or an Accrual basis.

If you were to account for revenue on a cash basis, revenue from each customer would be recognized when the money received from them is banked. This is typically pretty simple - we usually would ignore small differences in timing caused by credit card payments and cheques, recognizing sales at the time the sale transaction is settled. On this basis revenue is essentially recognized when the cash register rings or invoices are produced.

Accrual basis, on the other hand, attempts to match the recognition of revenue with the costs that relate to generating it. This can cut two ways: it is possible to defer revenue (recognize it in the future) when payment is received in advance of services being provided, for example, an annual credit card fee that covers the period May 1 - April 30 could be spread over the year it relates to. It is also possible to accrue revenue (recognize it in the present) when services are provided in advance of payment, for example, revenue for an ongoing contract might be taken into income monthly even though progress billings and payments by the customer are on a quarterly cycle, paid after the work is done.

In some industries this is a big deal. In construction, franchising and banking for instance, there are special rules indicating how to treat certain items of revenue laid down by accounting authorities. The accounting rules for financial reporting have a bias towards conservatism, however, that may be inappropriate for your customer profitability model. For example, accounting rule makers are very keen on deferring revenue but they are less inclined towards accruing revenue (since that can be used to inflate profit).

As a general rule accrual accounting provides a smoother recognition of profit in any given period, and is desirable for any purpose other than management of cash flow or business valuation. Most companies will choose to adopt an accrual basis of accounting for customer revenue. As you go through analysis of the types of revenue that occur in your business along with Finance you need to make sure that the treatment of revenue is as good an approximation of the business as is possible - even if it differs from your Finance policies. Remember that you are creating a management information measurement not a financial reporting system.

Cost

Costing is always a thorny issue. In most companies we see there are concerns that costing is not accurate enough, is not complete or is stale. This in itself is not a show-stopper when you want to build a customer profitability measurement. Most leading companies which have implemented customer profitability measurements do so using what costing information they have available, and then identify areas for future improvement, working with Finance to set the agenda for costing work. This can actually be a big win for the Finance function, as it drives relevance and direction (and even funding) in the process of costing.

Costs in a customer profitability model may include transaction-specific amounts or customer-specific amounts. Transaction-specific costs would include such things as the cost of a product purchased, opening an account or conducting a sale. Customer-specific costs may include such items as the cost of producing a customer statement, direct mail marketing or credit approval. Indirect costs are things like corporate overheads that are hard to relate to customers or transactions.

The ABC's of costing

When Finance prepares costing information, they usually use a method called Activity-Based Costing (ABC). In that process they identify activities which combine to produce products or services (often both internal and external). Using time and motion studies and analyzing the use of resources, they determine the cost of performing an activity, then determine the cost of various products and services.

When preparing ABC, it is common for costs to be identified as direct (caused by the activity, like cost of goods sold) or indirect (required for the activity to happen, like the cost of rent in a store) and also to have some fairly large chunks of cost dollars that remain unallocated to activities. It is also common to distinguish between fixed costs (those that do not change substantially with sales volume, like rent) and variable costs (which do, like cost of goods sold).

One of the decisions you need to make is whether to include all of these components. Direct variable costs are useful when making decisions at the margin (next customer, next unit) such as a negotiated price decision. For most product and customer management decisions, you will likely want to have a fuller inclusion of costs, usually direct and indirect, including both fixed and variable components. These less direct costs and particularly those of a fixed nature will require allocation to customers on some basis that will not distort the ranking of customers. Inclusion of unallocated costs is needed if you wish to be able to reconcile back to the Financial records of the company - a requirement in many implementations of customer profitability.

So what to do? We believe it is a good idea, when practical, to include both fixed and variable direct, indirect and allocated costs in the customer profitability model. Best practice is to store these levels of cost inclusion as separate subtotals of total cost, so that reporting can be based on a level of cost inclusion appropriate to the decisions which are being considered.

Accrual versus cash

As with revenue there are issues concerned with timing associated with cost, and the decisions you make will be affected by the same logic; it is generally desirable to match revenue and expense regardless of when they actually happen. Again, there is a bias among the accounting authorities towards conservatism that you should be aware of. Accounting authorities are keen on accruing costs but they are less inclined to permit deferral (since that can be used to inflate profit). In general, accrual which matches revenue and expense flows as closely as possible is what we should want in our customer profitability model.

Bases of costing

There are other issues which we also have to look at for costs. Over the years we have developed a number of bases for measuring costs which are specifically useful for different circumstances. Determining the basis of costing to be used in your customer profitability model is important as it will have a significant impact on the actual results you produce. The three main costing bases you need to know about and consider are Actual cost, Standard Cost, and Average cost.

Actual cost measures the cost of actually performing a transaction, making a product, taking a phone call or some other activity. Costs on this basis are usually determined by taking the expenses of the company and allocating them to activities. Actual cost is helpful for understanding the real profit of customers, products etc. taking into account efficiencies resulting from variances in day to day operations. Unfortunately, actual cost data is hard to get on a timely basis, since it can only be determined after the books are closed. For customer profitability we usually need something that is available faster to reflect the competitive urgency of the decisions we need to make.

Standard cost measures the cost of an activity according to an attainable performance standard. Costs on this basis are determined by process engineers who identify how the activity should be done and the value of resources that should be consumed to do it. Standard costs do not change from month to month. Standard cost is helpful when volumes are highly variable, especially in a product launch situation, where it would distort profitability to reflect the actual cost base over a small volume of activity.

Average cost most often allocates actual costs (from an earlier accounting period) to a customer-related cost driver such as account without reference to the activity level of the individual customer. Typically this method is used when behavioural cost drivers (activity data) are not available. To some extent it is necessary to use average costs to deal with unallocated costs in a customer profitability model. In the absence of ABC costing, average cost is often used in first implementations. This method of costing is generally inadequate for customer value management purposes. In the retail banking industry in Canada, for instance, dramatic shifts in customer profitability have been noted when changing from average cost per account to a more behaviour sensitive measurement using historical or standard cost.

Probabilistic items

In many businesses it is not possible to foresee the amount of cost that may be incurred in the future because an event may or may not happen in the future. This is a widespread issue dealing with probabilistic things such as the likelihood of a car accident claim, a loan default, goods being returned for credit or a coupon being cashed. It crops up in almost any business.

The first thing we should consider if you have items like these is how important they are to the overall numbers. If the amounts involved are small, you may be better off ignoring them.

If you decide that you need to take them into account, the next decision is whether you want to attribute the cost to an individual directly (for example a mail-in rebate might be tagged to an individual) or whether you want to take a portfolio approach to the probability of the cost and spread it over the entire relevant customer base (for example, allowing for a percentage of credit card balances to go into default). This is a something that is very specific to your business and needs to be decided internally.

If you need to make provisions for probabilistic future costs you will need to identify both the expected amount of the cost (difficult for car accident claims!) and the likelihood that it will happen. Both of these factors can usually be determined with reference to historical data. Once you have determined the probability it will happen and the expected amount it will cost, you will need to translate that expected cost (e.g. probability X amount) into an amount that you wish to attribute to each measurement period (e.g. monthly cost = expected cost/ account lifetime in months), and determine a basis for allocating that amount to your customers.

An illustration

To illustrate this process consider loan losses in a bank. You know that you will have some losses and you know some customers are more at risk than others, but you don't know how much will be lost, when or on which specific loan customers. Unfortunately it is also a big dollar item so it can't be ignored. So what do banks do?

First they recognize that the amount at risk is specific to each individual loan, depending on the amount of the loan, its insurance status and any collateral that backs the loan up as security. This solves the problem of estimating, at the account level, how much could be lost.

The second step is to identify the probability of default. This is done using, among other things, credit risk scoring of individual customers. Other considerations like the type of property pledged as collateral and the type of loan and the payment history of the customer may also be considered. Banks mine their historical data to determine historical probability of default for their customers based on factors such as these.

The next step is to identify the periodic charge for expected loss that should be reflected on the account. This is done with reference to the life of the loan itself, to ensure that when all the periodic charges are added up it equals the amount at risk times the probability of default (the total expected loss). Since all of these factors are tied specifically to an individual customer and their account, there is a high degree of precision in the allocation of these costs to individual customers.

Loan loss estimation in a bank is both mature science and sophisticated because of the mission critical nature of the cost. In most industries a less sophisticated analysis (using a similar framework, however) will prove adequate for measurement of indeterminate items.

Mapping decisions to accounting theories

If you are going to implement a Customer Profitability Measurement solution in your company you really do need to understand some of the technical choices that have to be made in the process of defining exactly how it should work so that your IT professionals can give you a system which provides you with what you need.

When you have to make choices in construction of a customer profitability measurement model the first question you should ask is why you can't have both alternatives available to you. For example, there is no reason that a customer profitability system can't give you both cash and accrual bases of measurement simultaneously. Similarly you can have different bases of costing available for analysis - if you store sufficient detail and specify the requirements for the model properly. Best in class solutions for Customer Profitability enable you to report on many different bases by storing the detail at a level that allows you choice of what to include at the reporting end of the process rather than at the calculation end.

The key thing to remember in all of this is that the decisions that you as a business manager must make are what should drive the specification of how customer profitability should be measured. You need a clear understanding of your position on these measurement issues to guide the work of your Finance and IT partners to make the measurement model useful to you.

Tags: Strategy, Analytics