Why standard costing quietly stops being true

A standard cost is two numbers multiplied together: a quantity you expect to consume, and a rate you expect to pay for it. Both of them were true on the day somebody worked them out. Neither of them is a fact about the future, and both drift — separately, at different speeds, for different reasons.

The failure mode is not that a standard becomes wrong. It is that it becomes wrong slowly enough that the variance report keeps looking plausible, and the plant keeps making decisions on it.

The four drifts, in the order they usually happen

Input rates move first, and visibly. Raw material prices, freight, power tariffs, a wage settlement. This is the drift everyone notices, and the one most likely to be maintained — because purchasing sees it every week.

Norms move second, and invisibly. A yield improves because an operator found a better nesting layout. A yield degrades because the tooling is worn. A supplier changes the incoming material’s thickness by a hundredth and the scrap rate moves with it. Nobody updates the BOM, because nobody was asked to and nothing appears to be broken.

The overhead base moves third, and catastrophically. This is the one that hurts. A machine hour rate is overhead divided by expected machine hours, and the expected hours were set at a planned utilisation. Run at 45% when you costed at 70% and the same absorption rate under-recovers by a third — so every product looks cheaper than it is, right at the moment when the plant can least afford to underprice.

The product mix moves fourth, and quietly invalidates any allocation basis that assumed the old mix. A basis that was fair when you made three variants stops being fair when one of them is 60% of volume and the setups are all on the other two.

The tell: variances that stop being noise

A healthy standard produces variances that scatter around zero. Some months adverse, some favourable, no pattern.

The moment a variance is one-sided for three or four periods running, it has stopped measuring performance and started measuring the gap between your standard and reality. Every month adverse on material price does not mean purchasing is failing; it means the standard rate is stale. Every month favourable on usage does not mean the floor got good; it means the norm is loose.

This is worth stating plainly because the variance report is usually read the other way round — as a verdict on the department rather than on the standard. That reading gets somebody blamed for a spreadsheet.

The four variances actually worth maintaining

You can decompose cost variance a dozen ways. Four of them earn their keep in a manufacturing plant:

• Material price — rate paid against rate assumed. Purchasing’s number, mostly outside their control.
• Material usage or yield — quantity consumed against the norm at actual output. The floor’s number, and the one that pays for itself.
• Labour and machine rate — recovery against actual hours, which is where a bad utilisation assumption shows up first.
• Absorption — overhead recovered against overhead incurred. The reconciling item, and the one that tells you the base is wrong before the others do.

Anything beyond these four tends to be analysis for its own sake, unless somebody has specifically asked the question.

How often to revisit

• Rates: quarterly. More often for a volatile input — resin, copper, steel, energy. Some plants run those at a rolling average and hold the rest annually, which is a reasonable compromise.
• Norms: on change, not on schedule. A new tool, a new supplier grade, a routing change, a machine replacement. These are events, and they should trigger a review rather than wait for a cycle.
• Overhead absorption: annually, with a mid-year check against actual volume. If volume has moved more than about ten per cent from plan, re-base it rather than explaining the under-recovery for six more months.

What silently breaks it between reviews

Two things, both of them floor-level and both invisible to finance:

Substitution. An alternate raw material used because the primary was out of stock, never fed back into the BOM. The job completes, the cost is booked at the standard for a material that was not used, and nothing in the system knows.

Undocumented process change. A step gets combined, a cycle time drops, an operation moves to a different machine with a different rate. The improvement is real. The standard still describes the old route, so the favourable variance it produces is read as efficiency rather than as a stale routing.

Both are caught by the same discipline: what the floor actually consumed and actually did gets recorded against the job, and the standard is compared to that rather than to a plan.

What MANUPRIME does about it

Costing is standard against actual, per batch and per unit, built from what the job card recorded rather than from a separate costing entry. BOMs carry versions, alternates and wastage norms, so a substitution on the floor is a recorded alternate rather than an invisible one, and a norm change is a new version rather than an overwrite. Material, labour, machine and overhead sit as separate elements so the four variances above come out of the same data rather than needing a reconciliation.

The costing module is one of the screens we walk through — ask for a demo as the accountant who would use it.