Production Costs: Structure, Formula, OEE Impact & How to Cut Them

What are production costs?

Production costs are the total monetary input required to turn raw materials into finished goods ready for sale. In textbook form that is material + labour + manufacturing overhead. In a real plant it is a far messier sum that includes machine depreciation, energy, tooling, scrap, rework, quality inspection, changeovers, indirect labour, internal logistics and a share of the maintenance budget. The gap between the textbook number and the plant number is usually where margin is lost — and where most cost-reduction programmes fail because they attack the wrong line.

In 25 years of automotive and FMCG plants, the pattern repeats: the cost report on the controller's desk is clean and monthly; the cost reality on the shopfloor is noisy and hourly. Cost per part fluctuates by 10–20 % shift-to-shift, driven almost entirely by OEE. When a plant reports a single "production cost" number per product, it is reporting an average that hides the variance that matters most. Production cost is a distribution, not a figure.

This is also why production costs and OEE are not two separate KPIs to be optimised in parallel. They are two views of the same underlying system. Every percentage point of OEE lost in the Availability, Performance or Quality factor translates directly into additional unit cost — not approximately, not philosophically, but arithmetically.

The four cost classifications that actually matter

Most confusion in cost discussions comes from mixing up two orthogonal classification dimensions. Every cost has a variable/fixed characteristic and a direct/indirect characteristic. The combination determines which lever actually moves the number.

The trap most cost programmes fall into: they aim at variable-direct costs (material) because those lines are the largest and the cleanest to measure. But in mature manufacturing — automotive, FMCG, packaging — the variable-direct cost is often already close to its theoretical minimum. The real loss is hidden in fixed and indirect costs, which are absorbed across whatever was actually produced. Produce less (low OEE), and every unit carries a larger share of the same fixed overhead. That is the mechanism that links OEE to cost — and it is the mechanism most cost analyses miss.

The production cost formula — with the line that gets forgotten

The canonical manufacturing cost formula is straightforward, but most plants apply it per order rather than per good part. That single substitution changes what the number means.

Total Production Cost = Direct Material + Direct Labour + Manufacturing Overhead

Unit Cost = Total Production Cost ÷ Good Parts Produced

The second formula is where the honesty lives. If you divide by all parts produced (including scrap), you are hiding quality loss in the overhead. If you divide by good parts, the same scrap pushes unit cost up — correctly, because the customer only pays for good parts. Plants that report cost per "produced" part instead of cost per "good" part systematically understate their cost base by 2–8 %. That is exactly the margin they later wonder where it went.

Direct material is the biggest line, but in most mature operations it is also the one with the least headroom — specifications are fixed, suppliers are negotiated, yields are close to theoretical maximum. The addressable cost in real improvement programmes sits in lines 3–7. Those lines are almost entirely controlled by OEE.

Why production cost is an OEE story in disguise

The arithmetic linking OEE to unit cost is straightforward and uncomfortable. Fixed and semi-fixed costs — depreciation, salaried staff, plant overhead, a big share of maintenance and energy — are incurred whether the machine runs or not. They are absorbed across whatever good parts come out the other end. So every OEE point lost increases the cost per good part by a mathematically predictable amount.

This is why every serious cost-reduction programme in discrete manufacturing eventually routes through the shopfloor: not because OEE is a fashionable KPI, but because fixed-cost absorption is where the real money is. A plant that holds material cost constant and moves OEE from 55 % to 75 % does not save 20 % on one line — it saves 20–35 % on the total unit cost, depending on how large the fixed-cost share is.

Where the avoidable cost actually hides

The second honest surprise in most plants: the cost that is theoretically addressable looks very different from the cost that gets attacked in the monthly operations review. The table below shows the typical pattern across automotive, FMCG and metal-forming plants I have worked in.

The first and third rows are the ones that matter most. They are the biggest single share of avoidable cost, and they are exactly the rows where paper- and ERP-based reporting is blind. A plant that cannot distinguish planned from actual cycle time, or that logs "micro-stop" as a single reason code at the end of the shift, is making cost decisions without the two largest inputs.

The cost-reduction sequence that actually works

Over 25 years the same sequence produces the same results, and deviating from it produces the same failures. The pattern is not original; what is original is how consistently the wrong sequence is chosen first, usually because the later steps are more visible to senior management.

Step 7 is almost always where cost-reduction programmes want to start. It is the most visible, the most presentable to a board, and the most expensive. Starting there without steps 1–4 means automating a process that is still losing 20 % to invisible downtime — locking the loss into faster, shinier hardware. The Meleghy rollout (10 % fewer stops, 7 % higher output, 5 % higher availability in six months) and the Klocke case (12 % higher output, 8 % higher availability in three weeks) both followed the correct sequence: visibility first, automation later.

What an MES changes in the cost conversation

FAQ

What is the difference between production cost and cost of goods sold (COGS)?
Production cost is everything spent to manufacture goods in a period, regardless of whether they were sold. COGS is the portion of that production cost that was actually shipped to customers and recognised against revenue in the same period. The gap between the two lives in inventory. Plants with high WIP and finished-goods stock can run at high production cost while reporting flattering COGS, because the cost is sitting on the balance sheet rather than the P&L. This is one of the quieter reasons why cash-strapped plants can still report "acceptable" margins until the inventory correction hits — and it is why serious cost programmes always look at production cost and inventory turns together, not in isolation.

How fast does an OEE improvement show up in unit cost?
Faster than most finance teams expect. The mechanism is fixed-cost absorption: the same salaries, depreciation and overhead are now spread across more good parts, which is a mathematical effect visible in the next costing cycle. In the Meleghy case, the 7 % output improvement and 5 % availability improvement translated into measurable cost-per-part reduction within the first quarter after rollout. The more fixed-cost-heavy your plant is (automated, capital-intensive), the faster OEE gains flow to unit cost. For labour-intensive operations the effect is smaller but still direct. The slowest line to move is material cost — yield improvements usually take two to three quarters to stabilise.

Why does scrap cost more than the "scrap rate" suggests?
Because the reported scrap rate almost always understates the real cost. A 2 % scrap rate on material cost sounds small — until you add in the labour already spent on those parts, the machine time they occupied (which displaced good parts), the energy consumed, the rework effort where applicable, the downstream disruptions, and the opportunity cost of the capacity they wasted. Fully loaded, a 2 % physical scrap rate typically costs the plant 5–8 % of total production cost. Plants that only track material scrap rate are looking at a quarter of the picture. This is why every serious MES quality module correlates defects back to the specific machine state, alarm and order that produced them — the physical part is the smallest piece of the loss.

When is material cost negotiation the wrong first lever?
When material cost is already at 55 %+ of production cost and yield is close to theoretical. At that point, squeezing the supplier by another 2–3 % is a one-time gain that usually comes with quality risk — and the same 2–3 % can often be found in OEE-driven fixed-cost absorption at zero supplier risk. The honest hierarchy: first fix what you control (OEE, scrap, changeovers, energy per part), then negotiate with suppliers from a position of cleaner demand forecasting and lower volatility. Suppliers give better prices to predictable customers, not to desperate ones. I have seen at least a dozen programmes where the sequence was reversed — supplier pressure first, shopfloor second — and the supplier relationship damage outlasted the one-off saving by years.

How does energy cost fit into the production cost picture in 2026?
Energy has moved from a line-item nuisance to a first-order cost driver in several industries — metal forming, glass, chemistry, plastics — and the volatility of the last three years has made it a planning problem rather than a budgeting one. The operational lever is the same as for every other cost: visibility first. Most plants still report energy at the meter level, which tells you the plant total but not where it is consumed. Modern MES architectures capture energy per machine, per state (running, idle, setup, standby) and per good part. The usual finding is sobering: idle and standby consumption is 15–30 % of total energy bill, for zero value added. That is almost always the biggest single energy saving — and it costs nothing to extract except a proper measurement architecture.

Related: OEE · Machine Downtime · Cycle Time · Lead Time · SMED · TPM · MES · Production KPIs

About the author

Christian Fieg

Head of Sales at SYMESTIC. Six Sigma Black Belt with 25+ years in automotive manufacturing. Ran global MES and traceability programmes at Johnson Controls (900+ machines, 750+ users across China, Mexico, US, Tunisia, France, Russia) and Visteon. Author of "OEE: One Number, Many Lies". · LinkedIn