Customer selection is a resource-allocation decision

A customer score is not a resource-allocation decision.

The short answer is that customer selection asks where scarce acquisition, retention, and relationship- development capacity should go, not only which customer has the highest score. A score can summarize value, risk, or response. It does not show what the firm must invest, which alternative use of capacity is displaced, or which relationship objective is being pursued.

Bhatnagar, Maryott, and Bejou frame customer selection and prioritization as an optimal resource-allocation problem. Their argument connects customer acquisition, retention, relationship development, and Customer Lifetime Value rather than assigning each to a separate budget. It also implies differentiated customers and investment levels instead of uniform relationship-marketing spending. The evidence supports a portfolio boundary, not a ranking of any current customer base.

The allocation problem also has a prediction boundary. Malthouse and Blattberg (2005) evaluate Customer Lifetime Value forecasts out of sample and show how the 20-55 rule can misclassify future high- and lower-value customers. Gupta et al. (2004) connect customer value to retention, contribution margin, customer count, acquisition, and discounting assumptions. Selection therefore has to carry both the resource constraint and the uncertainty of the value estimate.

Why does an account score fail to execute commercial budget allocation?

Suppose a customer list is sorted by a value score. The top row may be a useful candidate for attention, but the score has not yet answered the allocation question. A firm still has to decide what kind of resource is available and what the investment is meant to change.

The final row depends on the others. A value estimate without a resource boundary is a description. A resource budget without a relationship objective is a rationing rule. Selection becomes a decision when the firm compares options under the same declared frame.

The table makes the opportunity cost visible. Option A may be attractive, but its resource need changes what else can be done. Option C has uncertain value, yet a small test may be justified if learning is the declared objective. Option D cannot be compared because revenue is not a complete value boundary.

Which four interrelated investment decisions define customer selection?

Bhatnagar, Maryott, and Bejou connect four activities that are often managed as separate programs: customer acquisition, customer retention, relationship development, and Customer Lifetime Value. The connection matters because a customer can move through more than one decision. Acquisition may create the relationship. Retention may protect it. Development may change its scope. Customer Lifetime Value may provide a longer-run economic lens. None of these objects automatically determines the next investment.

Treating them as connected does not mean using one number everywhere. It means asking whether the allocation logic can see the relationship between actions and value. A retention resource can compete with acquisition capacity. A development investment can affect future value but require expertise now. A customer-value estimate can change when service costs or time horizons change.

The author-owned rule is therefore simple: state the relationship objective before applying the score. If the objective is acquisition, the relevant comparison may concern expected fit, acquisition cost, and future relationship development. If it is retention, intervention economics and preventable loss may matter. If it is development, the work required to expand the relationship becomes part of the choice.

Why is uniform commercial spending an unintended resource allocation choice?

Uniform relationship-marketing spending can look neutral because every customer receives the same rule. But equal treatment still consumes capacity. It may be appropriate when the service promise is deliberately uniform, when variation is not observable, or when fairness is the declared objective. It is not neutral by definition.

The source’s framework implies differentiated customers and investment levels. That implication does not say that a firm should always concentrate resources on a small group. Malthouse and Blattberg (2005) add the warning that a value forecast can misclassify future customers, so concentration also carries a prediction risk. It says the allocation problem should make the difference visible. A low-investment customer may receive a self-service route. A high-investment customer may receive specialist attention. A test customer may receive a limited learning resource. Each choice needs a value and capacity boundary.

This is where the portfolio view improves on a list. A list orders objects. A portfolio compares them under a constraint. The constraint may be contact time, service capacity, specialist availability, or a declared financial envelope. Without it, “prioritization” can mean only sorting.

Where does the operational cost-to-serve boundary limit customer value?

Customer Lifetime Value and customer profitability can be useful inputs, but neither is self-defining. Customer value may include revenue, margin, service cost, acquisition cost, retention cost, development investment, and the horizon over which the relationship is evaluated. A comparison that includes only revenue can reverse when the cost boundary is widened.

The correct response is not to reject scores. It is to state what the score is for. A value estimate can screen a portfolio. A profitability estimate can inform intervention. A relationship-development score can identify a learning or expansion option. The allocation decision still needs the resource and alternative-use columns.

Keep three tests visible:

1. Is the value boundary the same across the options?
2. Is the resource need measured on the same basis?
3. Would the decision change if the next unit of capacity went elsewhere?

If the third question has no answer, the article is looking at a ranking, not an allocation.

What disciplined decision sequence should govern account selection and pruning?

Use this sequence before a customer prioritization rule is turned into a management claim:

1. Name the relationship objective: acquire, retain, develop, learn, or exit.
2. Define customer value, including the revenue, cost, and time boundary.
3. Identify the resource that is actually scarce.
4. Estimate the resource need of each customer option.
5. Record what alternative use of capacity is displaced.
6. Select the decision horizon and the outcome that will be reviewed.
7. Keep exceptions and uncertain rows visible instead of forcing them into a rank.
8. Recheck whether the rule changes a decision or merely changes an order.

This process does not create an objective ranking without data. It creates a more honest question: what does the organization choose to fund, protect, test, or leave alone?

Which three conventional customer selection claims must leadership reject?

First, do not say “the highest-value customers deserve all the investment.” The value estimate does not remove capacity limits, relationship objectives, or alternative uses.

Second, do not say “every customer should receive the same relationship-marketing treatment” as if uniformity were costless. Equal rules still allocate scarce work.

Third, do not say “customer selection is solved by Customer Lifetime Value.” Lifetime value is an input whose boundary must be declared. Selection is the decision about where the next resource goes.

For adjacent decisions, compare preferred customer treatment with the customer-lifetime-value forecast boundary.

Where are the empirical limits of customer selection modeling?

Bhatnagar, Maryott, and Bejou provide a relationship-marketing framing in which customer acquisition, retention, development, and lifetime value connect to optimal resource allocation. Malthouse and Blattberg provide an out-of-sample prediction and misclassification boundary, while Gupta et al. provide the assumption boundary for customer-based valuation. The three studies do not rank the owner’s customers or provide a universal allocation rule. The synthetic matrix is an author-owned translation. It shows the fields that must be visible before a customer score becomes a portfolio decision.