Profitability cases: how to find where the profit went

Why profitability is the case type you will almost certainly meet

If you sit three first-round cases, expect at least one to open with some version of "our client's profits are down and they do not know why". It is the most common case type at all three firms, and it is common for a reason that has nothing to do with laziness on the interviewer's part. Profitability is the cleanest available test of whether you can take a vague problem and break it into parts that can actually be measured.

It is also the case type that every other case type borrows from. A market-entry case ends in a question about whether the new market will be profitable. A pricing case is an argument about one branch of the profit tree. A due-diligence case asks whether the target's profit is real and durable. A cost-reduction case is a profitability case with the answer given away in the prompt. Learn to decompose profit properly and you have learned the arithmetic spine of most of the case library.

Which makes it the worst case type to be mediocre at. Interviewers see it constantly, so they have a very calibrated sense of what a good version looks like, and the gap between a candidate who recites a profit tree and one who uses it to hunt is obvious within four minutes.

Across the transcripts we mark, the failure mode is almost never structural knowledge. Nearly everyone can name revenue and cost. The failure is sequencing: candidates start prescribing fixes before they have established what is broken. They hear "profits are down 40%" and within two minutes they are discussing whether the client should raise prices, cut headcount or launch a loyalty programme. None of that is an answer, because none of it is attached to a cause.

How do you build a profit tree properly?

Start from the identity, not from a remembered diagram. Profit equals revenue minus cost. Revenue equals volume times price. Cost equals fixed cost plus variable cost per unit times volume. Those three statements are definitionally true, and that is the entire source of the structure's authority.

This is worth being precise about, because framing is one of the four analytical sub-skills we mark and the thing it rewards is a structure that is genuinely exhaustive. A profit tree is mutually exclusive and collectively exhaustive because arithmetic makes it so. There is no third place profit can hide: it is either in what came in or in what went out. Contrast that with the four-box frameworks people memorise, where "customer", "competition", "company" and "context" overlap heavily and nothing guarantees the four boxes cover the problem. Those are checklists wearing a structure's clothing.

So say the identity out loud when you open. It takes eight seconds and it tells the interviewer that your tree is derived rather than recalled.

The second layer is where the case is actually won

The first layer is table stakes. Everyone gets to volume, price, fixed and variable. What separates candidates is having a real second layer ready, because the second layer is where a cause can be located precisely enough to act on.

One more thing about the tree: build it to fit the client. A subscription business wants revenue split into subscribers times revenue per subscriber, with churn under the first term. A retailer wants stores times sales per store. A manufacturer wants capacity utilisation somewhere in the cost branch. The identity is universal, the second layer is not, and framing marks whether you noticed the difference.

What order should you work through the tree in?

A tree is a map, not a route. Having drawn it, you need a sequence that eliminates as much as possible with each question. The following order does that, because each fork rules out roughly half the remaining tree.

The habit to build is saying the elimination out loud at each fork. "Cost per unit is essentially flat, so I am setting the cost branch aside and staying on volume" costs you four seconds and it is the single clearest signal of structured thinking available in a case.

What does a worked profit bridge look like?

Here is the arithmetic done properly on a small-appliance manufacturer whose profit fell from £80M to £48M in a year. The point of a bridge is that the pieces have to sum to the total. If they do not, you have either double-counted or missed a branch, and the check is free.

Read what that says. Volume is £20M of the £32M, so nearly two-thirds of the problem is units. Price and unit cost are £6M each: real, but secondary. This is a volume story, and saying so decisively is worth more than any further arithmetic.

Notice two things that catch people out. First, total cost actually fell, from £1,170M to £1,146M, while cost per unit rose. A candidate who only looks at total cost concludes that costs are under control and walks past a £6M problem. Cost questions are per-unit questions.

Second, look at the size of the volume move: units fell 4% and profit fell 40%. That is operating leverage. With £420M of fixed cost sitting under a £40 unit margin, small volume changes swing profit violently, which is exactly why the volume branch deserved to be found first.

So where do you look next? Not at pricing, and not at procurement. You ask for the 0.5M lost units split by product, channel and account, and you ask what the category did over the same period. Everything after this point is about explaining half a million units.

What separates a 3 from a 5 on a profitability case?

In our rubric, analytical thinking carries half the overall score and splits into framing, commercial judgement, rigor and synthesis. Scores run 1 to 5, where a 3 means you met the first-round standard rather than "average". On profitability cases specifically, the distance between a 3 and a 5 comes down to four habits.

Name a hypothesis before you ask for data

"Given the decline started when the new competitor launched, I would expect this to be a volume problem in the channel they entered. Can I see units by channel?" That sentence does three things at once: it commits, it explains the reasoning, and it makes the next data request purposeful. Compare it to "could I see the revenue and cost breakdown", which is a request for someone else to do the thinking. Interviewers can work with a wrong hypothesis stated clearly. They cannot mark a candidate who never has one.

Quantify the branch instead of asserting it

"Costs have risen" is an observation. "Cost per unit is up 50p, which on 12M units is £6M, so under a fifth of the £32M gap" is a finding. The second version tells you and the interviewer whether to keep pulling this thread. Rigor is marked on the auditable chain, so the number said out loud is worth more than the same number thought silently, and a branch you have not sized is a branch you cannot rank.

Check whether the market moved too

This is the single most reliable separator on this case type, and most candidates never ask. Take the worked example. Suppose the category fell from 100M units to 99M, a 1% decline, while the client fell from 12.5M to 12.0M, a 4% decline. Holding share steady in a 99M market would have given the client 12.375M units. So 0.125M of the lost units are the market shrinking, and 0.375M are share lost to somebody else. In profit terms that is £5M market and £15M share, which sums back to the £20M volume effect.

That split changes the recommendation completely. A £5M market problem is something you resize the cost base around. A £15M share problem is a competitive fight you either pick or concede, and it is where the client's own decisions are implicated. Calling one the other is the most expensive error available in this case type.

Use margins to test whether a price cut is affordable

When volume is the problem, the instinctive fix is to discount, and it is usually wrong. Do the arithmetic before you propose it. The client's contribution this year is 12.0M units at a £39 unit margin, which is £468M. Cut price by £5 and the unit margin becomes £34, so holding contribution flat needs £468M divided by £34, or 13.76M units. That is close to 15% more volume to pay for a 5% price cut. On a business that has just lost 4% of its units, assuming a 15% gain is not a plan.

How do you close a profitability case?

Synthesis is a distinct sub-skill and it is scored on a specific shape: recommendation first, the cause with its number attached, the risk, then the next step. Sixty seconds. No chronology of your own analysis, no "as I mentioned earlier", no list of everything you looked at.

The reason recommendation-first is non-negotiable is not stylistic. It is that a partner listening to you is deciding whether to act, and they need the verb before the evidence. A close that walks through revenue, then cost, then volume, then share, and arrives at a view in the last eight seconds has made the listener do the assembly.

Here is that close on the worked example, with the numbers we actually established.

Every number in that close was established earlier and none of them are new. That is the test of a good synthesis: it should be assemblable from your own notes without a single fresh calculation, which is why writing the bridge down as you go matters more than it feels like it does in the moment.

Two failure modes to name. The first is the hedge: "there are several drivers here and I would want to look at the data further". It reads as caution and marks as absence, and a reasoned view that turns out wrong genuinely outscores it. The second is the shopping list: five recommendations of equal weight, which tells the listener you never decided which cause was biggest. One recommendation, sized, with a risk you volunteered before anyone asked.