Cost Reduction Cases 2026: Find the Addressable Base

A cost reduction target is a dollar figure wearing a percentage

Here is the prompt. The figures are ours, posed as assumptions, and you should label your own out loud the same way.

Ardwick Industrial reports $400 million of revenue and $20 million of operating profit, so a 5 percent margin. Its cost base is $380 million. The board wants a 5 percent cost reduction.

Five percent of $380 million is $19 million. Note first what that would do: $19 million on top of $20 million of profit takes the margin from 5 percent to 9.75 percent. So the target is not a trim. It would nearly double profit, which is worth saying before anyone asks whether it is achievable.

Now find out what is actually reachable. Of the $380 million, $95 million sits under contracts with at least eighteen months to run: a leased distribution network, a software agreement, and a take or pay supply commitment. None of it moves inside the year.

Addressable cost is therefore $380 million minus $95 million, which is $285 million. Finding $19 million on $285 million requires 6.67 percent.

State the comparison exactly. The board asked for 5 percent of the reported base. Delivering it means cutting 6.67 percent of the part that can move, which is a third harder, since 6.67 over 5 is 1.333. That sentence is the answer to a question nobody asked, and it is why this case type rewards the denominator over the arithmetic.

Splitting the cost reduction base so the branches can be computed

A cost base split by accounting line produces branches nobody can act on. Split it by what governs the line instead, which is the convention Oliver Wyman publishes in its own fleet case, where it notes that "They do not expect costs to rise in a 1:1 proportion, because different types of costs are expected to change in different ways", checked 30 September 2026 at oliverwyman.com.

The column that matters is the third one. Two branches in this table cannot deliver a dollar in the current year no matter how good the analysis is, and a structure that treats all five equally has spent its time badly.

Read the second row carefully, because it is the one that gets mishandled. Direct labor is often called variable and behaves as a step cost: it moves when a shift is added or removed, and not smoothly with volume. A 3 percent volume fall does not produce a 3 percent labor saving, it produces nothing until a shift comes out.

The case interview step that doubles the answer if you skip it

Deloitte publishes a cost and benefit question that names its own two calculations, which makes it the clearest teaching artifact on this case type.

The client is a clothing and sport retailer. The question reads: "If Recreation Unlimited chooses to improve their website it will cost them $10M (one-time cost), but they expect that each visitor will spend an additional $1. The profit margin on any products purchased on the website is 50%. How would you evaluate the potential opportunity of improving the website? The client has also provided us with the data point they have 20M people visit the site annually", checked 30 September 2026 at usrecruiting.deloitte.com.

Then it defines both steps in its own words: "Calculation A: Contribution is the case influx generated by the additional visitors and spending. It does not account for the profit margin on products sold. Calculation B: Includes profit margin and is the actual profit generated." We quote that as printed; the answer table itself is an image, so the arithmetic below is ours.

Calculation A is 20 million visitors times $1, which is $20,000,000 of additional revenue. Calculation B is $20,000,000 times 50 percent, which is $10,000,000 of profit. The one-time cost is $10,000,000, so the investment returns its cost in one year exactly and every later year is clear profit.

Here is the trap, stated precisely. A candidate who reports Calculation A against the cost reports $20,000,000 of benefit against $10,000,000 of cost, which reads as a two times return. Calculation B puts the return at one times. Stopping one step early does not shade the answer. It doubles it, and the firm's own page names the step that was skipped.

When a published case interview does not reconcile

Arthur D. Little publishes a practice case about a gym chain, and its quantitative block does not close. The page reads: "It has 5 million citizens and a market size of 300 million EUR per year. Lux Flex hopes to cover 25% of the budget market and requires at least 210 million EUR of revenue per year to do so. With budget gyms' clients spending an average of 30 EUR per month (360 EUR yearly), they", checked 30 September 2026 at adlittle.com.

Twenty five percent of 300 million euros is 75 million. The page asks for 210 million, which is 70 percent of the same market. Both readings are arithmetically valid and they describe different businesses.

Work each one to members, since that is the operational number. At 360 euros a year, 210 million euros needs 583,333 members, rounding down to a whole person, which is 11.67 percent of a 5 million population. And 75 million euros needs 208,333 members on the same rounding, which is 4.17 percent. The ratio between the two is 2.8.

What to do in the room is the useful part. Do not silently pick one. Say that the stated share and the stated revenue requirement imply different figures, give both, and ask which the interviewer intends. A candidate who computes 583,333 without noticing has followed instructions. A candidate who notices has done the job, and this is a published exercise on a firm's own site rather than a hypothetical.

Price cuts in a cost reduction case, and the two numbers confused

Cost reduction cases often contain a hidden pricing question, and it carries the single most common arithmetic slip in the category.

Take a product at $50 with a variable cost of $38, so a contribution of $12, against $9.6 million of fixed cost. Break-even volume is $9,600,000 divided by $12, which is 800,000 units.

Cut the price by $2. Contribution falls to $10, and break-even volume becomes $9,600,000 divided by $10, which is 960,000 units. That is 160,000 more units.

Now the two figures that must be kept apart. The margin given up is $2 on $12, which is 16.67 percent. The additional volume required to stand still is 960,000 over 800,000, which is 20 percent. Both are correct answers to different questions, and calling the margin loss the volume requirement is the error.

The gap widens as margins thin. At a $6 contribution, a $2 cut loses 33.3 percent of margin and requires $9,600,000 divided by $4 against $9,600,000 divided by $6, so 2,400,000 against 1,600,000 units, which is 50 percent more volume. Thin margin businesses cannot discount their way to a cost problem's solution, and one line of arithmetic proves it.

Do not borrow sector slack as company slack

A tempting move in a cost case is to argue that spare capacity exists, so fixed cost can be spread over more volume. Check whose spare capacity it is.

The Federal Reserve reports August 2026 capacity utilization of 76.3 percent for total industry and 75.7 percent for manufacturing, against 1972 to 2025 averages of 79.4 and 78.2 percent, checked 30 September 2026 at federalreserve.gov. Individual lines vary widely in the same table: machinery at 83.4 percent, primary metals at 67.3 percent.

Those are weighted branch indexes. They say that slack exists somewhere in a branch of industry. They say nothing about whether a particular plant has any, and a case answer that treats the aggregate as evidence about the client has borrowed a number from a different population.

The honest version is short. Sector level slack is a reason to ask the client for its own utilization by asset, and the answer may well be that the constrained asset is running flat out while three others sit idle. That is a capacity mix problem with a different remedy from a cost per unit problem.

Sequence the cost reduction levers by when the cash arrives

A cost reduction recommendation is a sequence, and the sequence is set by timing rather than by size.

1. Discretionary spend. Travel, external advice, marketing tests. Small, and available inside a month.
2. Purchase price on direct materials. $168 million of base, so one point is $1.68 million. Available in a quarter with renegotiation.
3. Shift pattern. A step change, so it delivers a whole shift's cost or nothing. Weeks of notice, and reversible only at a cost.
4. Footprint. Site consolidation. Large, and it consumes cash before it releases any.
5. Contracted cost. $95 million of base, and nothing until the contract expiry date, which is why it goes last however large it looks.

Reading that list against the $19 million target tells you the answer is mostly rows two and three, and that the largest single block of cost is unavailable. Say that in the first thirty seconds of the recommendation.

Where the points go on a savings case

The scheme here is our own. The dimensions and their weights are read straight from the grading code in src/lib/grading, and because no consulting firm publishes a marking scheme, none of this is anyone else's standard.

What the sources do not settle

We looked for a first-hand candidate account of a cost reduction case and found nothing citable. Reddit refuses connections from this network, and the session had no video transcript tooling. Deloitte publishes a tips video and no account of a specific case. Arthur D. Little publishes two consultant interviews and neither discusses its practice case. Competitor preparation blogs cover cost cutting at length and are not citable sources.

Four things stay open. Arthur D. Little's own exercise does not reconcile, and we publish both readings rather than deciding for the firm. The Deloitte answer table is an image, so the calculation above is ours even though its inputs are the firm's. Ardwick's figures, including the $95 million of contracted cost, are ours and posed as assumptions. And no firm publishes a threshold for how deep a cut it will recommend, which is why this page sequences levers by timing rather than by an external benchmark.