Direct Material Usage Variance KPI

Direct Material Usage Variance is a critical KPI that measures the difference between the expected and actual costs of materials used in production.

It directly influences cost control metrics, operational efficiency, and overall financial health.

By tracking this variance, organizations can identify inefficiencies and implement corrective actions, ultimately improving ROI metrics.

A favorable variance indicates effective resource management, while an unfavorable one may signal waste or misalignment with strategic goals.

This KPI serves as a leading indicator for future performance, guiding decision-makers in their budgeting and forecasting processes.

How Direct Material Usage Variance Connects to Your Strategy

Direct Material Usage Variance holds two memberships in KPI Depot's database, and the two groups ask it to do different work. It sits in Cost Accounting, a group of thirty-four metrics, where it ranks eleventh. It also sits in Metals, a group of eighty-six metrics, where it ranks twenty-fourth.

The head of the Cost Accounting ordering runs Cost of Goods Sold (COGS), then Gross Profit Margin, Contribution Margin, Contribution Margin Ratio, Operating Expense Ratio, Variable Cost Percentage, Fixed Cost Leverage, and Break-Even Analysis. Every one of those states an outcome in money for a product line or for the business as a whole. This variance is the first metric in the ordering that points at a physical act on a shop floor. Its job in that group is explanation. When Gross Profit Margin or Variable Cost Percentage moves and nobody can say why, variance decomposition is where the answer is supposed to come from. Eleventh is an honest rank for that role. A controller does not put this metric in front of a board, a controller uses it to say what the board is already looking at.

Metals is ordered around physical production: Ore Reserves, Production Volume, Metal Recovery Rate, Yield, Cost of Production per Tonne, Energy Consumption per Tonne, Total Recordable Injury Rate (TRIR), and Lost Time Injury Frequency Rate (LTIFR). Six of those eight sit in the internal perspective. The role change is sharp. In Metals the physical fact this variance exists to detect, material consumed above what the process should have consumed, is already measured upstream and more precisely by Metal Recovery Rate and Yield, in physical units, by the people who run the line. So in a metals operation the variance is not a detector. It is a translator, converting a loss that metallurgy already knows about into a figure the ledger can carry. Twenty-fourth of eighty-six is the right rank for a translator.

Its balanced scorecard placement is the financial perspective, and that placement doubles as a warning label. The restatement is literal here in a way it is not for most financial metrics: the formula takes the quantity difference and multiplies it by standard material price, so the metric is a physical measurement wearing a currency symbol. Everything that determines it has already happened, and in the Metals KPI group the metrics that watched it happen are ranked well above it and sit in the internal perspective. Read it as a lagging reconciliation. It will never function as an early warning, because the warning was available in physical units first.

The tension worth naming first is with Material Price Variance, which the Cost Accounting group's own OKR material carries as a key result inside the same objective as this metric. The pair splits a single purchasing decision, and the split is decided by arithmetic rather than by cause. Because the quantity difference is valued at standard price, the entire price effect lands on price variance and the entire quantity effect lands here, whatever actually drove the outcome. Buy a cheaper grade that scraps more and price variance improves while this one deteriorates, on the same purchase order, with manufacturing absorbing the consequence of a decision it did not make. Anyone setting targets on both at once should decide in advance which metric owns the substitution case, because the metrics will not settle it between themselves.

In Metals the opposing metric is Production Volume, ranked second. Throughput is won by running faster, running hotter, and running whatever feedstock is in the yard, and each of those raises consumption per unit of good output. Cost of Production per Tonne, ranked fifth, conceals the trade rather than exposing it, since higher volume spreads fixed cost and can fall even while material loss climbs. The Cost Accounting group's OKR material supplies a second version of the same squeeze from the opposite direction: an objective built on Inventory Turnover Ratio and Days Sales of Inventory pushes toward leaner stock and smaller lots, smaller lots mean more setups, and setup scrap is charged to this metric. Lean inventory and material efficiency are not allies.

One last structural point about where this metric sits. Both groups file it as a measure of operations. It is equally a measure of the standard. Every other metric at the head of the Cost Accounting ordering is computed from transactions that actually occurred; this one is computed against an engineering assumption somebody wrote down, possibly years ago. The group's own best-practice material says as much, telling teams to recalibrate standard costing variances regularly so the cost model stays relevant. Until that recalibration runs on a schedule, any movement in this metric is ambiguous between the plant and the assumption, and the ambiguity is not resolvable from the number itself.

Measuring Direct Material Usage Variance in Practice

Settle what the metric is comparing against before anything else. The formula subtracts actual material quantity from standard material quantity and prices the gap at standard material price. Only one of those three inputs is an observation. Standard quantity is an engineering assumption: a bill of materials line, a routing, an assumed yield loss, written by someone who may no longer be in the building. So this metric measures the standard at least as much as it measures the shop floor, and a plant can run flawlessly against a wrong number and post a variance every period for years.

A Stale Standard Produces a Variance Nobody Can Close. If the bill of materials was written for a previous part revision, a previous supplier's coil width, or a machine that has since been replaced, the gap between standard and actual is structural. It reappears every period at roughly the same size, it survives every corrective action, and it is reported as a failure of people who cannot cause it. Two tests separate this case from a real one. A structural variance is stable in magnitude and sign across periods and roughly proportional to volume. A real one moves with lots, shifts, operators, or supplier lots. Before opening an investigation, check the age of the standard and the date of the last engineering change on the part. The fix is often a document revision rather than a shop floor action, and no amount of operator attention will substitute for it.

Standard Price Rescales the Whole Metric. The usage variance is a physical quantity multiplied by standard material price, which means the reported number moves when nothing physical has changed. Revalue the standard price at the annual cost roll and every prior period's usage variance is stated in units that no longer match the new ones. A plant that wasted exactly as much material as last year will report a different figure. For a volatile commodity input the rescaling can exceed anything the plant could have done operationally. Keep the physical quantity variance beside the valued one and publish both. The physical series answers whether the process improved. The valued series answers what it cost. In a year when the standard price was reset, the two can point in opposite directions, and only one of them is about the process.

Decide What Counts as Consumption. The formula says actual material quantity, and a plant has several defensible readings of that phrase. Scrap generated in process, material consumed in rework, samples pulled for inspection, destructive testing, startup and shutdown material, and material issued for engineering trials all draw on the same stock. Each can be charged into usage variance or booked to its own account. Both conventions are defensible and they produce different metrics. Booking destructive testing separately is usually the better choice, because testing volume is set by a quality plan rather than by process control, and leaving it inside makes a tightened inspection regime look like deteriorating discipline on the floor. Rework is the opposite case: it consumes material because something went wrong, so excluding it flatters the metric. Write the treatment of each category into the costing policy and publish that policy beside the figure. A customer comparing two plants without it is comparing accounting choices.

Inherent Yield Loss Versus Avoidable Waste. Some loss is physics. Trim on a coil, kerf on a cut, sprue and runner on a moulding, drying and evaporation loss, and metallurgical recovery limits are properties of the process and the part geometry, not of anyone's attention. The standard should already contain them, and where it does, the variance correctly reports only the avoidable remainder. Where the standard was set at theoretical yield, the variance reports physics as failure, permanently. That is the most common reason a usage variance runs chronically unfavourable in a cut-to-size or nested operation, and the correction is to set the standard at attainable yield for the actual material format and part geometry, then track the gap between attainable and theoretical as a separate engineering question with an engineering owner. Those are two different conversations and one number cannot host both.

Substitution Lands Here but Starts Elsewhere. When the specified material is unavailable and a different grade, gauge, or supplier lot is issued, consumption changes. A heavier gauge that meets the specification by being oversized consumes more mass per part. A lower grade scraps more. Either way the variance posts against production, while the decision was made in procurement, usually under a shortage that had its own urgency and its own justification. Capture substitution as a data field at the point of issue rather than as a remark in a variance meeting, so the period figure can be split into a substitution component and a process component. Without that field the metric quietly converts supply problems into manufacturing problems and hands the resulting number to the wrong manager.

Issue Timing Decides Which Period the Variance Falls In. The comparison assumes the material counted as used was used to make the output counted as produced. Backflushing at completion enforces that by construction, and in doing so it hides real consumption differences behind the bill of materials, which is to say it can suppress the very variance the metric exists to find. Issuing material to a work order as it is drawn does not enforce the match at all: material staged at a line and not yet consumed sits as an issue against an unfinished order, so an unfavourable variance appears in one period and an offsetting favourable one appears in the next. Long cycle times, campaign runs, and work in process held across a period end all produce this pattern. Either tie the period boundary to a physical work in process count, or report the variance on completed work orders only and accept the reporting lag that comes with it.

The Metric Inherits Every Inventory Error. Actual quantity is derived from stock movements, so an unrecorded issue, a return to store that was never keyed, a unit of measure conversion applied twice, a scale that drifted out of calibration, a theoretical consumption posted where a weighing was intended, and a cycle count adjustment dropped into a period end all arrive as variance. Nothing inside this formula distinguishes a record keeping error from a material loss, because both look identical: stock that left, output that did not appear. Publish inventory record accuracy for the relevant item classes next to this metric. Where record accuracy is poor, the variance is largely a measurement of the records, and it should not be used to direct shop floor action at all until that is fixed.

Batch Size Penalises Short Runs. Setup scrap, purge material, first-article pieces, and the length of strip lost threading a machine are fixed per run, not per unit. A short run carries the same absorbed loss over fewer good parts and posts a worse variance than a long run of the identical part made by the identical crew to the identical standard. An unsegmented figure will therefore rank small-lot cells, service parts, and new product introductions as the worst performers in the plant when their only distinguishing feature is lot size. Either build a setup allowance into the standard so that run length is neutral, or segment the reported figure by lot size band before anyone is asked to explain it.

Joint and By-Product Processes Break the Denominator. Where one input stream yields several saleable outputs, or a main product plus a by-product, there is no non-arbitrary statement of how much material a single output consumed. The allocation is a convention, whether by physical measure, by relative sales value, or at net realizable value, and the usage variance for any one output is a function of that convention. Change the convention, or let the relative prices of the outputs move, and the per-output variance moves with it while the process is untouched. For these operations the defensible measurement sits at the level of the whole process: total input against total saleable output, which is what Metal Recovery Rate and Yield already do in the Metals KPI group. Report a per-output usage variance only as an allocation exercise, and label it as one.

Where the Data Lives. The bill of materials and routing sit in the ERP, alongside a standard cost master that carries its own effective dates. Actual consumption comes from goods issue and material document history, production confirmations, and scrap declarations. Quality inspection lots hold the sampling and destructive test volumes. Physical inventory and cycle count history hold the corrections. The join that is usually missing is standard cost effectivity against transaction date: a variance computed with the current standard against last quarter's issues is not a variance, it is a revaluation. Rebuild the standard as of the transaction date rather than the reporting date, and keep the scrap reason code on every declaration, because a variance without reason codes can be sized but never explained.

Segment before aggregating, and in this order: by part or material family, then by lot size band, then by shift or line, then by whether a substitution was recorded. Keep the physical variance beside the valued one at every level.

The metric is at its best as a question generator on a narrow scope and at its worst as a plant-level score. Aggregated to a site it nets favourable and unfavourable movements from unrelated processes against each other and produces a small figure that means nothing at all, since the offsetting components can each be large. Held at part and work order level, with a fresh standard and trustworthy inventory records underneath it, it points directly at the thing that went wrong.

Common Pitfalls

Many organizations overlook the importance of regularly reviewing material usage data, leading to persistent inefficiencies that inflate costs.

• Failing to engage cross-functional teams can result in misaligned objectives. Without collaboration between procurement, production, and finance, discrepancies often go unaddressed.
• Neglecting to update forecasting models can skew variance analysis. Outdated assumptions may lead to unrealistic expectations, complicating the assessment of actual performance.
• Ignoring supplier performance metrics can mask underlying issues. If suppliers consistently deliver subpar materials, it can distort usage variance and impact production quality.
• Overlooking training for staff involved in material handling can lead to waste. Inadequate knowledge about best practices often results in inefficient use of resources.

Improvement Levers

Enhancing Direct Material Usage Variance requires a proactive approach to identifying and addressing inefficiencies in material management.

• Implement a robust inventory management system to track material usage in real-time. This allows for timely adjustments and reduces the likelihood of excess waste.
• Conduct regular variance analysis to identify trends and patterns. By understanding the root causes of discrepancies, organizations can implement targeted improvements.
• Foster strong relationships with suppliers to ensure quality and reliability. Engaging suppliers in performance discussions can lead to better material quality and reduced costs.
• Invest in employee training programs focused on efficient material handling. Empowering staff with knowledge can significantly reduce waste and improve overall operational efficiency.

KPI Depot is trusted by consulting, strategy, finance, and analytics teams at leading organizations worldwide, including those listed below.

OKRs That Use Direct Material Usage Variance

The Cost Accounting KPI group names this metric directly in its OKR material, which most KPIs in the database never get.

Drive operational efficiency through detailed variance analysis and control is the objective, and Direct Material Usage Variance is one of its key results, sitting beside Direct Labor Efficiency Variance, Cost Variance (CV), and Material Price Variance. The group's stated rationale is that variance analysis ensures operational inputs adhere closely to standards. That phrasing carries the whole risk of the objective, because it assumes the standards are right. A team can deliver every key result in it by tightening the shop floor. It can deliver them just as well by letting the standards drift toward actual consumption. The second route is faster, invisible in the metric itself, and it destroys the costing model the objective exists to protect.

The group's own best-practice material supplies the fix, and the fix belongs inside this objective rather than beside it. Two of the tips apply directly: recalibrate standard costing variances regularly so the cost model stays relevant, and review variances within the monthly close so deviations are caught while the evidence still exists. Turn both into key results. A directional set for this objective then reads: standards for the highest-consumption part families reviewed and re-dated within the cycle, a rising share of material variance attributable to a named and recorded root cause, a falling unexplained residual, and inventory record accuracy for production materials holding or improving. Those are the results that make a movement in the headline variance mean something. Without them, the objective can be satisfied by arithmetic alone.

The pairing with Material Price Variance in the same objective needs a rule written into the objective text. The two metrics divide one decision and the division is fixed by the formula rather than by responsibility, so a grade substitution improves one and worsens the other automatically, with no underlying change in performance. Either name procurement and manufacturing as joint owners of the substitution case, or set the two as a combined total material cost variance with the price and usage split reported underneath. Do one of those and the objective stops rewarding a transfer between two teams as though it were a saving.

The Metals KPI group takes a different line, and it is the right one. No Metals objective uses this metric as a key result. Its efficiency objective runs on Production Volume, Cost of Production per Tonne, Capacity Utilization, and Energy Consumption per Tonne, all physical or per-tonne measures owned by operations. Do not insert this variance into it. Its use in a metals setting is as the reconciliation that explains how a favourable Cost of Production per Tonne and a deteriorating Yield can be reported in the same quarter, and a reconciliation is a supporting analysis rather than a result someone is graded against.

One rule holds in either group. Freeze the standard for the life of the objective and say so in the objective itself. A variance target set against a moving standard is not a target, and the first person to notice that will not be the one who mentions it.

See OKR Examples for Cost Accounting

Free Whitepaper

Want to achieve performance excellence in Cost Accounting? Download our in-depth whitepaper: Definitive Guide to Cost Accounting KPIs.

Download the Free Guide

KPI Categories

This KPI is associated with the following categories and industries in our KPI database: