Estimate service cost with a time equation that reflects case complexity

An average cost per service case can hide meaningful differences in the work. A routine invoice inquiry may require a quick document check, while another needs a credit approval and records imported from an older system. A time equation estimates effort from those observable features without creating a separate costing category for every possible combination.

In a Harvard Business School interview, Robert Kaplan describes time-driven activity-based costing using the cost of resource capacity and the time demanded by transactions. He also explains how time equations add terms for features that change the work. The model below is an original finance-service example of that approach.

Choose a bounded process first. State where the case begins and ends, which team's effort is included and what the result will inform. A model for inquiry handling should not quietly become a measure of every cost associated with the customer relationship.

Suppose an illustrative inquiry process needs 4 minutes of base handling, 3 minutes for each supporting document, 12 additional minutes when credit approval is required and 6 additional minutes when an older-system lookup is needed. The estimated minutes are 4 + 3 × document count + 12 × approval flag + 6 × legacy flag.

Each flag is one when the condition applies and zero otherwise. A standard case with two documents and neither exception takes an estimated 10 minutes. A case with four documents, credit approval and a legacy lookup takes 34 minutes. These are invented time assumptions that would need observation in an actual process.

Use field definitions that can be recorded consistently. A document count should distinguish unique documents from repeated uploads. An approval flag should mean that the approval task actually occurs, rather than merely that the customer belongs to a category sometimes requiring one.

Assume the relevant team costs $54,000 for a period and supplies 60,000 practical working minutes for the modeled services. The capacity cost rate is $0.90 per minute. Under the simplifying assumption that all tasks use that same pool, the standard case receives $9 of cost and the complex case receives $30.60.

If the 12 approval minutes actually require a manager whose rate is $1.80 per minute, separate that resource. The complex case then uses 22 minutes at $0.90 plus 12 minutes at $1.80, totaling $41.40. A single blended rate would hide the cost of the different skill involved.

Keep the pool and capacity period aligned. State which employment, systems or premises costs are included and which available minutes support those costs. An allocated minute rate is not automatically the cash cost of an additional minute; the staffing commitment may remain unchanged when one case disappears.

A case open for two days may contain only twenty minutes of active work. Waiting for a customer reply should not be priced as continuous staff effort unless a resource is genuinely occupied throughout. Preserve processing effort and elapsed turnaround as separate measures because they answer different management questions.

Parallel work needs a different distinction. Two people each working for ten minutes consume twenty person-minutes even if the elapsed task takes ten minutes. Conversely, one employee running an automated job while doing another task should not automatically receive a full duplicated labor charge for both periods.

Observe representative cases across the features that drive complexity. A few convenient routine cases will not establish the time required for rare approvals. Record ranges and exceptions where the process varies, and test whether those uncertainties change the proposed customer or workflow decision.

The additive equation assumes each extra term represents work not already included elsewhere. If the base time was measured from cases that usually required a legacy lookup, adding another six minutes for that lookup may duplicate it. Define the reference case before estimating the increments.

Some features interact. A credit approval may require less additional work when the manager already reviewed the same documents for another reason. Alternatively, two exceptions together may require a separate reconciliation task. Add an interaction only when observation supports it and the difference matters to the decision.

Do not add dozens of terms to reproduce every past case exactly. The process owner must be able to maintain the field definitions and explain the resulting minutes. A small model with visible limitations can be more useful than a detailed equation whose inputs are rarely captured reliably.

Apply the equation to the complete selected case population and sum the estimated minutes by resource. Compare the total with the capacity actually supplied and with observed workload. A large mismatch can reveal missing tasks, inaccurate times, duplicate cases or a capacity definition that excludes important work.

Keep unused capacity visible under the chosen costing approach instead of automatically increasing every case's assigned minutes until the budget is fully absorbed. Likewise, estimated demand above available capacity requires an explanation such as overtime, backlog, subcontracting or an inaccurate model.

Sample cases with large predicted effort and cases with large differences from observed effort. Ask whether the cause is customer complexity, an internal correction or a process change. A cost assigned to a customer does not prove that the customer created avoidable work or should be charged for it.

A document-upload improvement might reduce repeated lookups, while a revised approval rule might reduce manager involvement. Change the affected term and apply it to the eligible case population. Then identify whether the resulting freed minutes support more work, shorter turnaround or a payment that can actually be removed.

Retain the time observations, equation version, rate calculation and case extract with the decision. After implementation, compare the relevant features and effort with the original assumptions. The model earns its place when it explains the work and supports a measurable change, rather than simply assigning a more elaborate cost to the same unexplained cases.