An operations / cost case is where cash leaks in the making and moving of things. The tree is not “cut 10% of everything.” It is: split fixed vs variable, locate the big line, then ask whether the leak is rate (too much $ per unit) or utilization (too little output on the same fixed cost).
Do not open a cement plant with 4Ps. Do not open it with a full value chain tour if the P&L already says energy is 38% of COGS — go there.
A tree that stays on cost
| Split | Question | Typical lever |
|---|---|---|
| Fixed vs variable | Does this cost move with tonnes this month? | Volume, make vs buy, shift structure |
| Rate vs usage | $ / unit vs units consumed | Energy price, mix, waste, speed |
| Utilization | Output / capacity | Uptime, bottleneck, campaign length |
| Make vs buy | Should this step exist here? | Contract mill, 3PL, outsource packing |
Profitability asks which of revenue or cost moved. This page assumes we already know cost is the problem (or the prompt is explicitly ops). If you have not isolated that, start with profitability.
Worked example: kiln energy, not “too many people”
Prompt. BalticCement’s coastal plant: 1.8 Mt clinker nameplate. EBITDA down $14m. Headcount is the CEO’s favorite villain. COGS breakdown:
| COGS line | $m | % of COGS |
|---|---|---|
| Energy (coal + power) | 41 | 38% |
| Raw materials | 28 | 26% |
| Logistics (inbound coal + outbound) | 24 | 22% |
| Labor (plant) | 12 | 11% |
| Other | 3 | 3% |
| COGS | 108 | 100% |
Labor is $12m. Even firing 20% of the plant would save ~$2.4m before you break the kiln crew and pay overtime. You cannot find $14m in labor. Stop talking headcount.
Energy rate vs usage. Last year 3.6 GJ / t clinker at $6.4 / GJ → $23.0 / t. This year 3.9 GJ / t at $7.1 / GJ → $27.7 / t. Output 1.41 Mt actual (was 1.52 Mt).
Energy cost this year: 1.41e6 × 27.7 ≈ $39.1m (plus power residual to match the $41m line).
Bridge the $ energy hit:
- Price: 3.6 GJ/t × $0.7/GJ × 1.41 Mt ≈ $3.5m
- Usage (heat rate): 0.3 GJ/t × $7.1 × 1.41 Mt ≈ $3.0m
- Volume down on energy is mostly variable — less tonnes, less energy $ — so volume is not why energy $ stayed high; rate is.
Utilization. 1.41 / 1.80 = 78%. Last year 1.52 / 1.80 = 84%. Fixed cash (labor, some maintenance) spread over fewer tonnes. Labor $12m / 1.41 Mt = $8.5/t vs $7.9/t last year: ~$0.9m. Real, not the story.
Heat rate 3.9 vs 3.6. Cause: more stop-starts (unplanned downtime 9% vs 5%) and wetter raw mix. Each cold start dumps energy. The ops tree says: uptime and kiln stability, not a hiring freeze.
Logistics 22%. Inbound coal is CIF. A cheaper coal with worse GJ/t can fake a “price win” and lose on heat rate. Check $ / GJ delivered, not $ / tonne of coal.
Recommendation. Do not cut 40 operators. Fund the maintenance that kills stop-starts (the $3m heat-rate leak plus reliability). Hedge or contract coal on $ / GJ. Accept that $3.5m of the hit is market price — that is a margin story for HQ, not a plant-manager moral failure. Risk: pushing uptime without spares causes a 3-week outage. Next step: cold-start count and coal GJ/t on one chart.
When this tree is the wrong tool
Revenue is the hole. Mix and price will not yield to a utilization speech.
The cost is mostly allocated overhead from HQ. Plant “savings” that move costs to another cost center are not savings. Push back.
The case is a value-chain position question (where to play in quarry vs retail). Use value chain. This page is inside the plant P&L.
The mistake unique to ops/cost cases
Equal slices of a pie chart. Candidates allocate 15 minutes to labor because it is a human story. Labor was 11%. The unique fail is hunting headcount when the COGS table already named energy. Interviewers put 38% on the page on purpose.
Second unique fail: cutting variable cost by cutting volume. Running the kiln less “to save coal” blows utilization and raises $/t on fixed cost. Say whether the lever is rate or run-rate volume.
How to open
“Cost is 80% energy, materials, and logistics. I will ignore a labor narrative until those three move. I will split energy into $ / GJ versus GJ / tonne versus uptime.”
Then the $3.5m / $3.0m / $0.9m bridge.
Isolate the cost line, then the rate
Structure drills on ops prompts should start from the biggest COGS line, not from a generic cost list.