Executive summary: Throughput accounting measures three numbers instead of allocating overhead: throughput (price minus truly variable cost), investment (money tied up in inventory and assets), and operating expense (everything else). It exists because traditional cost accounting rewards building inventory and misprices products, pushing plants to overproduce and to chase the wrong orders. This guide shows the three measures, how to rank products by throughput per constraint hour, and why that ranking routinely reverses the one your margin report gives you.
- What is throughput accounting?
- What are the three measures?
- What counts as truly variable cost?
- Why does cost accounting mislead manufacturers?
- How do you rank products by throughput per constraint hour?
- Which decisions change most under throughput accounting?
- How do you introduce it without abandoning GAAP?
- Throughput accounting: operator FAQ
- About the Stagnation Assassin
What is throughput accounting?
Throughput accounting is a decision-making framework that measures the rate a business converts sales into money rather than allocating overhead to units. It tracks throughput, investment, and operating expense, and it judges every decision by its effect on system throughput. It exists because allocated-cost thinking systematically rewards overproduction and misprices work.
The framework came out of Theory of Constraints, developed by Eliyahu Goldratt alongside the operational method. The insight behind it is uncomfortable but simple. Cost accounting was built for a world where direct labor was the dominant cost and where making more units genuinely made you money. Neither is true in most plants now. Direct labor is often a small fraction of cost, overhead dominates, and making more units than your constraint can absorb actively destroys value.
Here is what makes this practical rather than academic. I have led transformations at Berkshire Hathaway, Illinois Tool Works, and Whirlpool, and in almost every one, the finance system was quietly arguing against the operational fix. The plant needed to produce less at non-constraints. The cost system rewarded producing more. The plant needed to take a specific order at what looked like a thin margin. The cost system said reject it. You cannot win that fight with operational logic alone, because at month end the numbers decide who was right.
Throughput accounting is how you stop the finance system from fighting the operations system. It does not replace your statutory books. It sits alongside them as the decision lens, and it usually reverses somewhere between a quarter and a third of the calls a standard margin report would have you make.
What are the three measures?
Throughput is revenue minus truly variable cost, the rate money enters the business. Investment is money tied up in inventory, equipment, and assets. Operating expense is all money spent turning investment into throughput. Every decision is judged by whether it raises throughput, lowers investment, or lowers operating expense, in that priority order.
Throughput
Throughput is selling price minus the costs that genuinely change with each unit produced and sold. Not allocated overhead. Not absorbed labor. Only what varies unit by unit. This is the money the business actually generates, and it is the measure with unlimited upside, which is why it ranks first.
Investment
Investment is money the business has tied up: raw material, work in process, finished goods, equipment, and facilities. Inventory sitting on your floor is not an asset in this framework. It is cash you cannot spend, and it earns nothing until it becomes throughput.
Operating expense
Operating expense is everything spent to turn investment into throughput: salaries, rent, utilities, depreciation, indirect labor. Note that this includes most of what cost accounting spreads across units. Here it stays visible as a period cost rather than being buried into inventory value.
The priority order matters more than the definitions. Throughput first, because it has no ceiling. Investment second, because releasing tied-up cash is fast and finite. Operating expense last, because it is the most limited lever and the one most likely to damage the business if you cut into muscle. Most struggling companies do this in exactly the reverse order. They cut operating expense first, which is why so many cost-cutting programs produce a good quarter and a worse year.
What counts as truly variable cost?
Truly variable cost is only what changes with each additional unit sold: raw material, purchased components, sales commissions, and outbound freight. Direct labor usually does not qualify, because you pay the same wages whether the line runs one unit or a hundred. Getting this line right is what separates throughput from ordinary contribution margin.
This is where most first attempts go wrong, and the error is almost always the same one. Teams classify direct labor as variable because that is how the standard cost sheet has always treated it. Ask the honest question instead: if we make one more unit today, does our labor bill change? In a salaried or fixed-shift operation, it does not. Those people are paid whether the unit gets made or not, which makes their cost operating expense, not truly variable cost.
The exception is real overtime and genuine piece-rate work, where an extra unit genuinely triggers an extra dollar of wages. Include that. Exclude the rest. The same test applies everywhere: does this cost move when one more unit moves?
- Almost always variable: raw material, purchased components, consumables destroyed per unit, sales commission, outbound freight, per-unit royalties
- Almost never variable: supervision, rent, depreciation, utilities, quality staff, maintenance salaries, indirect labor, allocated corporate overhead
- Depends, so test it: direct labor (variable only if genuinely triggered per unit), energy on high-draw equipment, tooling with per-unit wear
Being strict here has a consequence people find alarming at first. Your throughput per unit will look far larger than your standard margin, often by a wide gap. That is not optimism. It is the removal of costs that were never caused by the unit in the first place. Those costs are still real and still have to be covered, but they are covered in total by total throughput, not pretended into each unit through an allocation formula.
Why does cost accounting mislead manufacturers?
Absorption costing capitalizes overhead into inventory, so building units the customer has not ordered raises reported profit. It also allocates overhead per unit, which makes low-volume products look unprofitable and high-volume products look better than they are. Both distortions push exactly the behavior constraint management is trying to stop.
Let me take the inventory distortion first, because it is the more dangerous of the two. Under absorption costing, overhead is absorbed into the value of inventory produced. Produce more than you sell, and a chunk of this period’s overhead moves onto the balance sheet inside inventory value instead of hitting the income statement. Reported profit goes up. Nothing about the business improved. You converted cash into product nobody ordered and booked it as performance.
Now put that next to what a constraint-managed plant is trying to do. It is deliberately running non-constraints below capacity so they produce only what the constraint can consume. Under absorption accounting, that shows up as unfavorable volume variance and lower reported profit in the transition period. So the operations team does the right thing and gets punished for it at month end. I have watched this exact collision derail good implementations more than once. It is why companies destroy value while believing they are creating it.
The allocation distortion is subtler. Spreading overhead by labor hours or machine hours produces a per-unit cost that has nothing to do with what the unit actually consumed. A product that barely touches your constraint but takes many labor hours looks expensive. A product that monopolizes constraint capacity but is quick elsewhere looks cheap. You then price and prioritize off those numbers, and you end up filling your scarcest resource with your least valuable work.
Under absorption costing, producing 1,000 units against 800 units of demand raises reported profit, because overhead rides onto the balance sheet inside inventory value. Nothing improved. The plant converted cash into unsold product and booked it as performance, while the constraint-managed decision to build only 800 shows up as an unfavorable variance.
How do you rank products by throughput per constraint hour?
Divide each product’s throughput per unit by the constraint minutes it consumes, then rank. Because the constraint sets system output, the scarce resource is constraint time, not revenue or margin. This single calculation frequently reverses the priority order your margin report produces, and it is the highest-value use of the framework.
Work a concrete case. Two products, one shared constraint.
Product A: sells for $500, truly variable cost of $300, so throughput is $200 per unit. It consumes 10 minutes of constraint time. That is $200 divided by 10 minutes, or $20 per constraint minute, which is $1,200 per constraint hour.
Product B: sells for $900, truly variable cost of $400, so throughput is $500 per unit. It consumes 40 minutes of constraint time. That is $500 divided by 40 minutes, or $12.50 per constraint minute, which is $750 per constraint hour.
Every margin report in the building says Product B is the winner. Higher price, higher absolute margin, better percentage. Sales is compensated to chase it. And it is the worse product, by sixty percent, measured on the only resource that is actually scarce. Every constraint hour you spend on B instead of A costs the company $450 of throughput it will never recover.
This is the calculation that changes how a company sells. Once you rank the full portfolio this way, three things usually fall out immediately. There are products you should raise prices on because they eat constraint time disproportionately. There are products you should push hard because they generate exceptional throughput per constraint hour. And there is usually a tail of work that consumes real constraint capacity while generating almost nothing, which belongs on a kill list.
Which decisions change most under throughput accounting?
Four categories flip most often: whether to accept a low-priced order, whether to make or buy, whether to run a batch larger than demand, and where to spend improvement money. In each case cost accounting asks what the unit costs, while throughput accounting asks what happens to system throughput, and the answers frequently diverge.
Accepting a low-priced order
A customer offers a price below your fully loaded unit cost. Cost accounting says reject it, you would lose money on every unit. Throughput accounting asks two questions instead. Does the price exceed truly variable cost, and does the order consume constraint time? If it clears variable cost and uses spare non-constraint capacity, it is pure additional throughput and you take it. If it consumes constraint time that better work could use, you decline it regardless of how the margin looks.
Make versus buy
Outsourcing decisions built on fully loaded costs are wrong more often than not, because most of the loaded cost does not disappear when you outsource. The rent and the supervisors remain. The right test is whether outsourcing frees constraint capacity that can be filled with throughput-generating work worth more than the outsourcing premium.
Batch sizing
Cost accounting rewards long runs because they spread setup cost across more units, lowering apparent unit cost. Throughput accounting looks at what the extra units actually are: inventory, tied-up cash, and carrying cost, unless demand exists for them. Long runs are justified at the constraint, where setup time is system capacity. Everywhere else they mostly manufacture working capital problems.
Where to spend improvement money
A cost-reduction project on a non-constraint reduces a number on a report and changes nothing about what the company earns, because system output did not move. The same money spent at the constraint converts nearly one for one into throughput. Throughput accounting makes that difference visible before the money is spent instead of after.
An order priced below fully loaded cost is profitable whenever it clears truly variable cost and runs on spare non-constraint capacity. The same order is a mistake if it consumes constraint hours that higher throughput work could fill. Cost accounting cannot tell these two situations apart, because it never asks what the order does to the constraint.
How do you introduce it without abandoning GAAP?
You run it as a parallel decision lens, not a replacement for statutory reporting. Keep GAAP books for external reporting, and build a throughput view for internal decisions. Start with one constraint and one product family, prove the ranking changes real outcomes, and expand from there. No accounting standard is violated by thinking clearly internally.
This is the objection that stops most CFOs, and it is worth answering directly. Nobody is suggesting you stop complying with GAAP. External reporting requirements do not change. What changes is which numbers you use to decide what to build, what to price, what to accept, and where to invest. Companies run parallel internal views for all sorts of purposes already. This is one more, and it happens to be the one that tells you the truth about capacity.
A practical sequence that works. Identify the constraint properly first, because throughput per constraint hour is meaningless without knowing what the constraint is. Then calculate truly variable cost for one product family, applying the strict test rather than copying the standard cost sheet. Rank that family by throughput per constraint hour and compare it to the current margin ranking. The gap between those two lists is your business case, and it is usually stark enough to end the debate.
Then pick one decision, ideally a pending order or a pricing review, and run it both ways in front of the leadership team. Show what cost accounting recommends, show what throughput accounting recommends, and show the throughput consequence of each. I have never had to do this more than twice before a leadership team asked for the full portfolio ranked the new way. The arithmetic does the persuading, which is a relief, because the change management on everything else in constraint work is considerably harder.
One caution. Do not let this become a finance project run inside finance. The value shows up when sales knows which products to push, when operations knows which orders protect the constraint, and when the improvement budget follows constraint impact. If it lives in a spreadsheet the operators never see, it produces interesting analysis and no throughput.
Throughput accounting: operator FAQ
What is throughput accounting in simple terms?
It measures the rate a business converts sales into money using three numbers: throughput (price minus truly variable cost), investment (money tied up in inventory and assets), and operating expense (everything else). Decisions are judged by their effect on system throughput rather than on allocated unit cost, which prevents the overproduction that cost accounting rewards.
How is throughput different from contribution margin?
Contribution margin usually treats direct labor as variable. Throughput excludes it unless an additional unit genuinely triggers additional wages, which in most fixed-shift operations it does not. That stricter definition matters because it stops labor absorption from making overproduction look profitable and gives a cleaner basis for ranking work by constraint time consumed.
Why does cost accounting encourage overproduction?
Absorption costing capitalizes overhead into inventory value, so producing more units than customers ordered moves overhead onto the balance sheet and raises reported profit. Nothing improved, but the numbers say otherwise. Meanwhile a plant correctly producing only what its constraint can absorb shows an unfavorable volume variance for doing the right thing.
Does throughput accounting replace GAAP?
No. It runs as a parallel internal decision lens while statutory reporting continues unchanged. Keep GAAP books for external reporting and use the throughput view to decide what to build, price, accept, and improve. Start with one constraint and one product family, prove the ranking changes outcomes, then expand.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B+ in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as VP of Global Product Strategy. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via the speaking page or connect with him on LinkedIn.
Next step: a throughput ranking
Your margin report is almost certainly telling you to sell the wrong products. Book a 20 minute throughput review and I will help you rank your portfolio by throughput per constraint hour, then show you which products to price up, which to push, and which are quietly eating the capacity you cannot buy back. Start the review here.