One rate was doing two jobs, and doing neither well.

The challenge

Work center rates are the hinge of a manufacturing cost system. Get them wrong and every product cost, every quote, and every make-or-buy decision inherits the error. This business had them wrong in a specific and consequential way: one rate structure was being asked to do two incompatible jobs.

As a pricing instrument the legacy model worked. It produced consistent, market-referenced quotes and gave the commercial team predictable customer outcomes. As a costing instrument it distorted almost everything it touched.

The same rate cannot simultaneously reflect what production actually consumes and what the market will bear. Asked to do both, it will quietly optimize for neither.

Three distortions followed. It compressed the margin curve across work centers, so genuinely different operations looked similarly profitable. It masked real cost differences between factories, which made multi-site decisions guesswork. And most damagingly, it disincentivized capital investment. Buy a machine that improves throughput, and the allocated rate on that work center falls, making the investment look worse in the very system meant to evaluate it.

What the numbers were actually made of

The starting position, once examined, was thinner than anyone assumed. Labor rates were identical for every work center at every plant save two international sites. Fixed overhead was calculated per plant. And the variable overhead rate was not calculated at all. It was a plug, derived by subtracting labor and fixed overhead from a total departmental rate. The residual was carrying whatever the model could not explain.

The build

Bifurcate the model

The recommendation was to stop asking the legacy model to be two things. A modern, data-driven costing model would reflect what each work center actually consumes. The legacy rates would be retained deliberately as a pricing tool, where their consistency and market grounding are a genuine commercial asset.

That distinction matters in a way that is easy to miss. Most cost-transformation work assumes the old model is simply wrong and must go. Here the old model was well suited to one of its two jobs. Naming which job, and keeping it there, preserved commercial discipline while fixing the cost side.

Direct labor, from three systems

Labor cost per hour at each work center was assembled by joining three sources that had never been joined: hourly pay rates from the HR system (anonymized by employee file number), the employee-to-work-center-to-factory mapping from the ERP, and actual, earned, and machine hours from the shop floor execution system.

Real data has holes, so the calculation was built with a deliberate fallback hierarchy: use the work center rate where it exists; fall back to a department weighted average by hours; fall back again to a factory weighted average. Nothing silently defaults to zero, and every fallback is visible.

Overhead, split and allocated on the right driver

The overhead pool was taken from the monthly cost walk that ties to the plant income statements, all cost accounts less direct labor, then categorized as fixed or variable rather than left as a residual. Fixed overhead allocates on square footage per work center, provided by factory leadership. Variable overhead allocates on hours. Two different drivers, because they are two different behaviors.

Reconciled to the income statement

The model was run against a year-to-date period and reconciled three ways: income statement cost against actual hours, calculated cost against earned hours, and both against the prior plugged rate. A costing model that cannot be walked back to the financials is a spreadsheet with opinions in it.

What it changes

How to tell if you have this

Three questions, and you will know within a few minutes.

• Is your variable overhead rate calculated, or is it what remains after labor and fixed overhead are subtracted from a total? If it is a residual, it is absorbing every error upstream of it.
• Do two work centers with visibly different equipment, labor, and floor space carry the same or similar rates? If so, the model is averaging away the thing you need it to show.
• When a work center takes on faster equipment, does its allocated rate fall in a way that makes the investment look worse? If yes, your costing system is arguing against your capital plan.

The Margin Diagnostic is scoped to answer these against your actual data rather than your assumptions about it.