What standard costing assumes
Standard costing sets an expected cost per unit from an expected set of inputs, then reports variances against it. In discrete assembly this works well: a product needs four of one component and two of another, and it needs those every time.
Batch manufacturing violates the premise. A run may yield 940 kg where the recipe predicted 1,000. An input lot may be 3% more concentrated than the last, changing how much is needed. Neither is an error to be explained away - it is the normal behaviour of the process.
The three costs that move
When a batch costs more than expected, the cause is almost always one of three things, and separating them is what makes the number actionable.
- Input price - the same materials cost more than the standard assumed, which is a purchasing question
- Input quantity - more material was consumed than the recipe called for, which is a process or formulation question
- Yield - the same inputs produced less output, which is where batch manufacturing loses money most quietly
Landed cost is not optional here
For manufacturers importing inputs, freight, duty, and handling can be a material share of the true cost of a lot. Costing that uses invoice price alone understates it, and the gap is not evenly spread - it lands hardest on dense, low-value, or heavily regulated materials.
Because those costs attach to a specific delivery, they attach naturally to a lot. Spreading them across an item average instead loses the very variation you are trying to see.
Cost per lot, not cost per item
Two lots of the same item bought three months apart can have genuinely different costs. Averaging them into one item cost is convenient, and it hides the thing you most want to know: whether your margin is drifting because of purchasing, process, or yield.
Costing at lot level also makes valuation explainable. When finance asks why inventory is worth what it is, the answer is a set of specific lots at specific costs rather than a rolling average nobody can decompose.
What to do with the variance
A variance number that nobody owns is decoration. Each of the three causes has a different owner, and routing them accordingly is what turns costing into a management tool.
Price variance belongs to purchasing. Quantity variance belongs to production or formulation. Yield variance usually belongs to process engineering - and it is the one most likely to be quietly absorbed rather than investigated, because it appears as a smaller number spread across many batches.