On this page
Salesforce size is a response model, not a headcount ratio. The right question is which constraint a change in size is meant to relieve: missing coverage, overloaded service work, weak prospecting, slow information gathering, or low productivity in an existing assignment.
That framing prevents a familiar category error. A ratio compares people with accounts or revenue as if the denominator described the work. It does not. The same number of people can produce a different result when territory potential, information quality, travel, selling effort, or management control changes.
Why is optimal salesforce size a market response model rather than a capacity ratio?
Beswick and Cravens connect force size, effort allocation, territory formation, forecasting, and performance standards in one analytical model (Beswick & Cravens, 1977). Size is one decision in a sequence. A change to it can alter the effort available to each assignment, but its effect depends on how the assignment and response function are represented.
Horsky and Nelson make that dependency operational with an efficient-frontier approach. Their method compares district productivity with alternative allocations and force sizes (Horsky & Nelson, 1996). In two firm applications, the larger opportunity often lay in productivity improvement in weaker districts rather than simply adding people (Horsky & Nelson, 1996). That is a decomposition, not proof that every business should improve before hiring.
Three counterfactuals before one answer
When a team says it needs more people, write three rows before approving the response.
Figure 1The headcount counterfactual
Compare the response to the constraint. A headcount change is not a diagnosis until the evidence says which output or service burden it changes.
Source: Author's decision worksheet grounded in Beswick and Cravens (1977), Darmon (2002), Horsky and Nelson (1996), and Piercy, Cravens, and Morgan (1999).
More people can leave the information bottleneck intact
Darmon treats customer-information work as time that competes with selling (Darmon, 2002). That gives the headcount question a useful failure mode. If each additional salesperson inherits the same poor information, the new capacity may spend its first period learning the assignment rather than creating more selling time. The constraint has been replicated, not relieved.
This does not make information work a defect. Learning a customer base can be the necessary work. The point is to name it. A hiring request that says “more coverage” may actually be a request for better account records, clearer qualification, faster access to product knowledge, or more time for prospecting after service obligations are met.
The model’s limits matter. Darmon’s numerical application uses one mechanical-parts supplier and 25 salespeople, and its later recommendation was not implemented (Darmon, 2002). It demonstrates how information enters the capacity problem. It does not establish a general force size.
Productivity is not a moral judgement
Frontier benchmarking can make a weak district look like a person problem. That is not what the method establishes. It compares observed responses with a constructed frontier and asks where the gap might be associated with productivity, allocation, or size (Horsky & Nelson, 1996). A lower result can reflect assignment potential, information quality, competitive conditions, or a mismatch between behaviour and the outcome being judged.
Piercy, Cravens, and Morgan separate behavioural performance, outcome performance, and organizational effectiveness in a cross-sectional survey (Piercy et al., 1999). That separation is a useful guard against single-number management. More activity can be evidence of effort and still fail to show that the organization received the outcome it needed. A capacity decision should say which of those objects is expected to change.
What to do with the three rows
First, write the constraint in observable terms. “We need more people” is not one. “The assigned service load leaves no protected time for new-account work” is closer. “The territory has potential that current coverage cannot visit within the required interval” is closer still.
Second, choose the smallest comparison that could separate the responses. Reallocation might move one workload unit. Improvement might remove one information handoff. Addition might cover one unserved segment. The comparison needs a defined output and a time window, otherwise every response can claim credit for movement that would have happened anyway.
Third, name the result that would stop the response. If a new salesperson does not change coverage or selling time, the original constraint was misdescribed. If reallocation moves activity but not the outcome, the assignment was not the whole problem. If process improvement releases time but the pipeline does not respond, the missing constraint may be demand or proposition fit.
Start with a capacity definition
Headcount is easy to count because it is a visible input. Selling capacity is harder because it is what remains after service, travel, information work, coordination, and learning have consumed time. The Beswick and Cravens architecture links size to effort allocation and territory formation rather than treating it as a stand-alone ratio (Beswick & Cravens, 1977). A capacity request should therefore state which kind of time is missing.
More coverage can mean that some accounts receive no attention. More selling time can mean that existing people are carrying too much service or information work. More expertise can mean that the response function is constrained by a capability rather than by the number of people. Those statements may lead to different interventions, even when the proposed solution is initially the same hire.
The definition need not begin with a precise local number. It can begin with the work that the new capacity is supposed to release or create. Without that sentence, a headcount plan can be approved while the original constraint remains unchanged.
The three responses fail differently
Adding capacity can fail when the new person inherits the same weak information, territory design, or service burden. Reallocating capacity can fail when the boundary moves but the underlying demand does not. Improving the work system can fail when time is released but the market or proposition does not provide a responsive opportunity. These are different failure modes, not reasons to reject any one response.
Horsky and Nelson’s frontier approach separates productivity improvement, geographic reallocation, and force-size changes in two firm applications (Horsky & Nelson, 1996). The separation gives a useful counterfactual discipline. Ask what each response would change, what it leaves untouched, and which observation would distinguish it from the other two. A response that cannot be distinguished cannot teach the organization whether the problem was capacity, allocation, or productivity.
The comparison also protects against moral language. A weak result is not automatically a weak person. It may be a mismatch between the assignment and the response function, a learning burden that was not budgeted, or an output measure that does not represent the work being asked for.
How should commercial organizations measure true salesforce response?
The result of a capacity change should be specified before activity is counted. More calls, visits, or opportunities may be useful intermediate signals, but they are not interchangeable with service coverage, qualified demand, retained relationships, or commercial outcomes. Piercy, Cravens, and Morgan distinguish behavioural performance, outcome performance, and organizational effectiveness in their work on territory design and control (Piercy et al., 1999).
That separation gives a practical measurement order. First record the behaviour the response should make possible. Then record the outcome that behaviour is expected to influence. Finally record the organizational effect that matters to the decision. If the first row changes and the second does not, the response may have relieved time without solving demand. If the second changes and the third does not, the local gain may not justify the system cost.
The order is not a causal model. It is a way to stop one favourable activity number from carrying all the meaning of a headcount decision.
A reversible headcount case
Write the case in five lines before requesting a permanent addition. Name the constrained work, the capacity response, the competing reallocation and process responses, the observation window, and the stop condition. Use neutral units if the case is being shared outside the organization. Do not add a private account list simply to make the request look concrete.
The case is stronger when the new person has a defined job in the model. Is the role intended to cover unserved potential, absorb service work, improve information quality, or create a specialist response? Those are not interchangeable. The role should also have a result that could show the constraint was elsewhere.
If the evidence remains ambiguous, the correct conclusion may be to postpone the hire while collecting the missing observation. That is not indecision. It is a decision to buy information before buying capacity. A headcount request earns approval when the organization can say what will change, what will not, and how it will know the difference.
Boundary
The research supports a decision architecture that keeps size, allocation, productivity, information, and performance distinct. It does not support a universal sales-rep ratio, a causal reading of frontier status, or a claim that hiring is usually wrong. Headcount becomes a sensible response when the constraint it relieves is explicit and the alternative counterfactuals have been given a fair test.
The response question connects to the asset that can leave and the territory as a workload model, because team size is meaningful only against the capacity and relationship mechanisms it is meant to carry. When individual effort is difficult to observe directly, variable pay is a risk design when effort is hard to observe, linking expansion decisions to contractual incentive structures.
References
- Beswick, C. A., & Cravens, D. W. (1977). A multistage decision model for salesforce management. Journal of Marketing Research, 14(2), 135-144. https://doi.org/10.1177/002224377701400201
- Darmon, R. Y. (2002). Salespeople's management of customer information: Impact on optimal territory and sales force sizes. European Journal of Operational Research, 137, 162-176. https://doi.org/10.1016/S0377-2217(01)00049-2
- Horsky, D., & Nelson, P. (1996). Evaluation of salesforce size and productivity through efficient frontier benchmarking. Marketing Science, 15(4), 301-320. https://doi.org/10.1287/mksc.15.4.301
- Piercy, N. F., Cravens, D. W., & Morgan, N. A. (1999). Relationships between sales management control, territory design, salesforce performance and sales organization effectiveness. British Journal of Management, 10, 95-111. https://doi.org/10.1111/1467-8551.00113
