In a typical first-round ops cost case, five of eight candidates open with a flat 10% cut across every COGS line. Two of those five still pass on structure. None of them sound like people I would put in front of a plant manager. The interviewer is not scoring thrift. They are scoring whether you can name the cost driver that actually moves the P&L.
That is the whole game. Split fixed from variable, ask whether the plant is underutilized, then spend your airtime on the one line whose rate or usage explains the EBITDA hole. Everything else is decoration.
After this page you should be able to open an ops cost case in under a minute, build a three-line bridge the interviewer can follow, and recommend a lever that does not quietly raise unit cost by starving volume.
What manufacturers actually put in the 10-K
Real filings do not hand you a case-style pie chart. They tell you which costs move with volume and which do not.
Caterpillar's 2024 Form 10-K defines variable manufacturing costs as those with a direct relationship to production volume: material, direct labor, freight, power to run machines, and supplies consumed in the process. Period manufacturing costs support production but generally do not move with short-term volume: machinery repair, depreciation on manufacturing assets, facility support, procurement, scheduling, and operations management.
Ford's 2024 Form 10-K is blunter on the fixed side: a high proportion of relatively fixed structural costs means small changes in wholesale unit volumes can significantly affect profitability. Components and raw materials sit on the other side. Soft volume with an untouched structural base raises unit cost whether or not anyone "cut costs."
U.S. GAAP pushes the same idea into inventory accounting. Under ASC 330, variable production overhead is allocated on actual use of the facilities, while fixed production overhead is allocated on normal capacity. You do not jack up the per-unit fixed overhead charge just because the plant ran light; the unabsorbed fixed cost hits the period. In a case, that is why "cut volume to save money" is often a trap: you save variable spend and strand the fixed base.
The four questions that replace a generic cost tree
I stop listening when a candidate lists seven cost categories and then visits them in alphabetical order. Use a decision sequence instead.
| Question | What you are testing | Typical lever if yes |
|---|---|---|
| Does this cost move with units this month? | Fixed vs variable | Volume, make-vs-buy, shift pattern |
| Is $/unit rising, or are we buying more units of input? | Rate vs usage | Price, mix, yield, waste, energy intensity |
| Is output / nameplate capacity soft? | Utilization | Uptime, bottleneck, changeover, demand |
| Is cash stuck in WIP while the line waits? | Flow time | Batch size, release policy, queue at the constraint |
The first three cover most first-round prompts. The fourth shows up when the case mentions inventory, lead time, or a bottleneck that is "always busy" while finished goods are late. That is where Little's Law earns its keep.
John Little's 1961 result in Operations Research is L = λW: average items in a system equal arrival (or throughput) rate times average time in the system. In plant language, WIP = throughput × cycle time. If WIP is up and throughput is flat or down, cycle time lengthened — a flow problem, not a headcount problem. The RePEc abstract of Little (1961) is enough primary grounding for an interview.
Capacity utilization is the macro sibling. As of August 2026, the Federal Reserve's G.17 release put manufacturing capacity utilization at 75.7%, about 2.5 points below its 1972–2025 average. Plants running light absorb fixed cost over fewer units.
Worked example: Northline Corrugated
Prompt. Northline runs one 110-inch corrugator and three converting lines. EBITDA is down $9.2m year over year on roughly flat revenue. The CEO's narrative is overtime and "too many people on third shift." You get this plant P&L for COGS.
| COGS line | This year ($m) | Share | Last year ($m) |
|---|---|---|---|
| Linerboard and medium | 62.0 | 54% | 58.4 |
| Plant labor (incl. overtime) | 19.5 | 17% | 18.1 |
| Energy (gas + power) | 13.8 | 12% | 12.9 |
| Adhesives, ink, starch | 6.1 | 5% | 5.8 |
| Maintenance and tooling | 8.4 | 7% | 7.2 |
| Other plant | 5.2 | 5% | 4.9 |
| Total COGS | 115.0 | 100% | 107.3 |
Shipments: 780 million square feet (MMSF) this year vs 850 MMSF last year. Nameplate on the corrugator: 1,050 MMSF on a realistic three-shift calendar. Converting WIP average: 42 MMSF this year vs 28 MMSF last year. Average order cycle time from corrugator release to ship: 6.1 days vs 3.8 days.
Labor is the story the room wants. Labor is not where $9.2m lives.
Step 1 — Rank the money, ignore the narrative. Paper is $62m. A 5% yield swing on paper is about $3.1m. A 20% cut to plant labor is about $3.9m before you break the crew that runs the corrugator. You can debate labor later. First explain paper and volume.
Step 2 — Split paper into rate and usage. Board cost per MMSF:
- Last year: 58.4 / 850 = $68.7 / MMSF
- This year: 62.0 / 780 = $79.5 / MMSF
Bridge the $10.8 / MMSF rise on this year's volume:
- Price / mix on board: interviewer gives +$4.0 / MMSF on the same basis weight mix → 4.0 × 780 = $3.1m
- Usage (yield): remaining 6.8 / MMSF × 780 = $5.3m
Yield is the leak. Scrap rose from 6% to 11% of board into the corrugator, driven by more short-run changeovers (average run length fell from 48 MMSF to 29 MMSF). Each changeover dumps warm-up waste and out-of-spec flute. Candidates who "negotiate board price" leave $5.3m on the table.
Step 3 — Utilization on the fixed base. Corrugator utilization: 780 / 1,050 = 74% this year vs 850 / 1,050 = 81% last year. Period costs that barely moved with volume (maintenance crew base, depreciation in "other," salaried supervisors inside labor) still sit near last year's dollars. Rough plant fixed cash of ~$28m spread over 70 fewer MMSF is about $2.5m of unit-cost damage even before overtime. That matches Ford's structural-cost warning in miniature: volume soft, fixed base sticky, $/unit up.
Step 4 — Little's Law on converting. Throughput fell from 850 to 780 MMSF per year. WIP rose from 28 to 42 MMSF. Using WIP = TH × CT on a 250-ship-day year:
- TH last year: 850 / 250 = 3.40 MMSF/day; WIP 28 → CT ≈ 28 / 3.40 ≈ 8.2 days
- TH this year: 780 / 250 = 3.12 MMSF/day; WIP 42 → CT ≈ 42 / 3.12 ≈ 13.5 days
The case's stated corrugator-to-ship cycle time (3.8 → 6.1 days) is the same direction on a tighter system boundary. Either way, WIP grew while throughput fell. The plant is not short of people. It is releasing too many short jobs onto converting, building queue, and starving the corrugator of long campaigns that protect yield.
Recommendation. Do not cut third-shift converters first. Raise minimum run length or batch same-flute orders to cut changeovers; that attacks the $5.3m yield hole and lifts corrugator utilization. Accept ~$3.1m of board price as a commercial problem for procurement and sales, not a plant moral failure. Overtime is a symptom of thrashing through short runs. Risk: longer runs raise finished-goods inventory for SKUs with lumpy demand — check SKU-level turns before you freeze the schedule. Next chart: changeovers per week vs scrap % on one page.
Ned's rule. If a COGS line is under 20% of cost and the CEO's favorite villain, spend sixty seconds killing the narrative and move to the largest variable line. Interviewers put the big number on the exhibit on purpose.
Where candidates leak points
Equal airtime. Fifteen minutes on labor because it is a human story when paper is 54% of COGS. The scoring sheet rewards prioritization, not empathy for the org chart.
Cutting variable cost by cutting volume. Idling a shift "to save board and energy" strands the fixed base and can raise scrap further if restart quality is poor. Say whether your lever is rate, usage, or run-rate volume.
Confusing allocated HQ overhead with plant cash. Moving depreciation between cost centers is not a plant win. Ask what cash leaves the company.
Skipping flow when the exhibit shows WIP. If inventory and lead time are in the packet, run a thirty-second Little's Law check.
This page assumes cost is already the problem. If you have not isolated revenue vs cost, start with a profitability case before you open an ops tree. If the question is where in the chain to play, use a value chain frame instead of a plant P&L.
How I want you to open
"Cost is mostly board, labor, and energy. I will ignore a headcount story until yield and utilization explain the EBITDA gap. I will split board into $ per MMSF versus MMSF consumed, then check corrugator utilization and converting WIP."
Then the bridge: $5.3m yield, $3.1m board price, ~$2.5m utilization drag. That is how you sound like someone who has sat in a weekly ops review.
Drill the bridge, not the vocabulary
Take any ops cost prompt and force a three-number bridge in five minutes: price, usage, volume or utilization. CoachNed's math drills are built for that pace — short reps with instant scores, no slide theater.
Everything on CoachNed is open for seven days with no card; then $120 for a recruiting season or $49 a month. If you want a full voice case after the drill, use /interview. CoachNed is not affiliated with McKinsey, BCG, Bain, Deloitte, or any firm named in this article.
Practice this today
Open /drills/math, pick an ops or profitability numerical, and write only: (1) the largest COGS line as a share, (2) a rate-vs-usage split on that line, (3) utilization = actual / nameplate. Stop when you have a one-sentence recommendation that names the driver, not a laundry list of cuts.
Frequently asked questions
What is the operations cost framework in a case interview?
It is a prioritization tree, not a checklist: fixed vs variable, rate vs usage, utilization, and — when WIP or lead time appear — flow via Little's Law. Find the cost driver that moves EBITDA; do not trim every line by the same percent.
How do you separate fixed and variable costs in a case?
Ask whether the cost changes if volume rises 10% this month. Materials, energy to run equipment, and overtime labor usually do. Depreciation, salaried supervisors, and much of maintenance scheduling usually do not. Caterpillar's 10-K uses the same split for variable vs period manufacturing costs.
Why does capacity utilization matter for unit cost?
Fixed costs spread over fewer units when utilization falls, so $/unit rises even if cash variable spend falls. As of August 2026, Federal Reserve G.17 manufacturing utilization sat below its long-run average — the macro version of that plant math.
When should you use Little's Law in an ops case?
I stop listening when a candidate lists seven cost categories and then visits them in alphabetical order. Use a decision sequence instead.
When the exhibit shows WIP, queues, or cycle time alongside throughput. WIP = throughput × cycle time. Rising WIP with flat or falling throughput means flow time lengthened — usually scheduling or a bottleneck, not a blanket headcount cut.?
Take any ops cost prompt and force a three-number bridge in five minutes: price, usage, volume or utilization. CoachNed's math drills are built for that pace — short reps with instant scores, no slide theater.
Is cutting headcount ever the right answer in a cost case?
Sometimes, if labor is the large variable line and volume has permanently reset. In the Northline example, labor was 17% of COGS and not the $9m hole. Lead with the math share, then decide.
Sources
- Caterpillar Inc. Form 10-K for the year ended December 31, 2024 — variable vs period manufacturing cost definitions used in the fixed/variable section. Checked 2026-09-24.
- Ford Motor Company Form 10-K for the year ended December 31, 2024 — fixed structural costs and volume sensitivity; raw materials vs structural cost framing. Checked 2026-09-24.
- Federal Reserve G.17 Industrial Production and Capacity Utilization (current release) — manufacturing capacity utilization 75.7% as of August 2026 vs long-run average. Checked 2026-09-24.
- Little, J. D. C. (1961), "A Proof for the Queuing Formula: L = λW," Operations Research (RePEc abstract) — primary statement of Little's Law used in the flow section. Checked 2026-09-24.
- ASC 330-10 inventory costing (normal capacity and fixed overhead) — fixed overhead allocated on normal capacity; unabsorbed fixed cost expensed when production is abnormally low. Checked 2026-09-24.