Product Launch Case Interview 2026: Cannibalization

Cannibalization decides a product launch case interview

Four branches cover a launch. Demand and willingness to pay, the price and the margin it leaves, the route to market, and the cost to produce and serve. Those four are fine and they are also what every candidate says.

The fifth branch is the one that carries the decision. Some share of the new product's volume will come from the company's own existing line. Those units were already earning a margin, so the company pays for them twice: once in the cost of the launch, and once in the margin it gives up on the line the launch replaces.

Put the fifth branch in the structure at the start. It changes the objective function rather than adding a consideration, which is why appending it at the end of a case reads as a rescue. The clean formulation is that a launch is worth its gross contribution minus the contribution displaced from the existing portfolio minus the fixed cost of launching.

Oliver Wyman publishes what it scores on exactly this kind of question. On the quantitative step of its pricing case it lists the "Ability to identify and ask for key information needed, for example the size of the addressable market" ahead of the "Ability to perform calculations and quantitative analysis", checked 30 September 2026 at oliverwyman.com. Asking for the cannibalization rate is the single most valuable information request available in a launch case, and the firm's own criteria reward asking before computing.

Cannibalization as a subtraction, in one worked launch case

Here is the case, with every figure stated so it can be rebuilt.

A snack manufacturer sells a bar to retailers at $1.10 with $0.72 of cost, so unit contribution on the existing line is $0.38. It has developed a protein bar to sell at $1.50 with $0.95 of cost, so unit contribution on the new line is $0.55. The year one forecast is 40 million units, and the fixed cost of launching is $14 million covering moulds, listing fees and launch media.

Gross contribution first. 40 million units at $0.55 is $22 million.

Now the subtraction. The commercial team estimates that 35 percent of new bar volume will come from buyers who would otherwise have bought the existing bar. 35 percent of 40 million is 14 million units, and 14 million units at $0.38 is $5.32 million of contribution that the existing line stops earning.

Net incremental contribution is $22 million minus $5.32 million, which is $16.68 million. Against $14 million of fixed launch cost the launch clears $2.68 million in year one.

Notice how thin that result is. A $22 million gross number became a $2.68 million net number, and the whole reduction came from two lines the forecast did not carry. A recommendation built on the gross figure would have described a launch nine times more attractive than the one on the table.

The launch stops paying above a 52.6 percent rate

A single point estimate of the cannibalization rate is the weakest part of the case, because nobody in the room knows it. So stop defending the estimate and solve for the value that matters.

The launch breaks even when net incremental contribution equals the $14 million fixed cost. That leaves $22 million minus $14 million, which is $8 million of room for displaced contribution. Each cannibalized unit costs $0.38, so the launch can afford $8 million divided by $0.38, which is 21.05 million cannibalized units.

Expressed as a rate, 21.05 million divided by 40 million is 52.6 percent. The launch survives up to a cannibalization rate of 52.6 percent and fails above it.

That is the sentence to lead the recommendation with. It converts an argument about an unknowable input into a testable threshold, and it tells the client what to measure. A regional test that reports source of volume gives the number directly, and a test that reports only incremental sales does not.

The threshold also shows where the sensitivity lies, and it is not where candidates expect. Move the existing line's unit contribution from $0.38 to $0.50 and the affordable units fall to $8 million divided by $0.50, which is 16 million units, so the breakeven rate drops to 40 percent. A company with a high margin incumbent has less room to cannibalize itself than a company with a low margin one. That result reverses the usual intuition that strong existing products make launches safer.

A declining incumbent buys 17.6 points of tolerance

One correction turns the arithmetic from defensible into right, and it is almost never applied.

Cannibalization costs money only when the displaced unit would otherwise have survived. Suppose the existing bar has been losing 8 percent of its volume a year to competitors. Part of the 14 million units the new bar takes were going to leave the franchise anyway, and the margin on them was already forfeit.

Credit a quarter of the cannibalized volume as volume that was leaving. Then 14 million times 0.75 is 10.5 million units genuinely displaced, and 10.5 million units at $0.38 is $3.99 million. Net incremental contribution becomes $22 million minus $3.99 million, which is $18.01 million, and the year one result rises to $4.01 million.

The breakeven rate moves with it. The launch can now afford $8 million divided by $0.38 of genuinely displaced volume, which is still 21.05 million units, but those represent 21.05 divided by 0.75, so 28.07 million units of apparent cannibalization, which is 70.2 percent of launch volume. The decline rate of the old line buys 17.6 points of tolerance.

This is the branch that separates a strong launch answer from a competent one. A declining incumbent makes a launch more attractive, and the mechanism is precisely that the margin being cannibalized was already draining away. Saying that out loud, with the two numbers, does more than any qualitative point about portfolio strategy.

Price the launch first, because price beats volume

Price and volume arrive in a launch case as separate inputs, and treating them separately hides the most useful finding in the whole case type.

Reprice the protein bar at $1.80 instead of $1.50. Unit contribution becomes $0.85, and the commercial team's demand curve says volume falls to 26 million units. Gross contribution is 26 million units at $0.85 is $22.1 million, which is effectively the same $22 million as before.

Now apply the same 35 percent cannibalization rate. 35 percent of 26 million is 9.1 million units, and 9.1 million units at $0.38 is $3.458 million. Net incremental contribution is $22.1 million minus $3.458 million, which is $18.642 million.

So the higher price is worth $18.642 million against $16.68 million, a gain of $1.962 million, on identical gross contribution. The mechanism is clean: cannibalization is charged per unit, so the option that moves fewer units displaces less margin. At equal gross margin, price beats volume in any portfolio with something to cannibalize.

Oliver Wyman's published pricing case shows how wide the plausible range can be. It reaches a cost based floor of INR 15,000 and a value based ceiling of INR 100,000, concluding that "the product can be priced within a range of INR 15,000 to INR 100,000", checked 30 September 2026 at oliverwyman.com. That spread is 6.7 times. At the floor, one million users at INR 15,000 produce INR 15 billion against INR 10 billion of fixed cost, so the margin is 33.3 percent. At the recommended top of INR 30,000 the revenue doubles to INR 30 billion and the margin reaches 66.7 percent. Same product, same cost base.

What three firms publish on product launch, and what they withhold

McKinsey publishes the cleanest launch breakeven in the market. Electro-Light launches "at a price of $2 to retailers", each bottle "would cost $1.90 to produce and deliver in the new process", and SuperSoda "needs to bear $40 million in total fixed costs", checked 29 September 2026 at mckinsey.com. Unit margin is 10 cents, so breakeven is $40 million divided by $0.10, which is 400 million bottles. At one eighth of a gallon each that is 50 million gallons.

The case then names "the target 12.5 percent share of the electrolyte-drinks market". For 50 million gallons to represent 12.5 percent, the category has to be 400 million gallons. That is an implication of the firm's own numbers rather than a figure published on the page, and the exhibit that would confirm it is an image we have not read. Say which it is when you use it.

ZS publishes the only consumer launch case with a worked profit answer, and its treatment of cannibalization is the lesson. The answer reads "They serve distinct segments, so cannibalization is low. Product B is premium, priced four times higher", checked 30 September 2026 at zs.com. Then the profit calculation runs from 140,000 buyers to "about $2.1 million in annual profit" with no unit subtracted for the low cannibalization the same answer asserted. Low and zero are different numbers.

Oliver Wyman's case has a gap of a different kind, and it is worth knowing because a candidate will hit it. The page states 10 million autistic children in India and an income split of 50, 40 and 10 percent, then works from a user base of 4.2 million. Middle and high income together give 5 million. Nothing published on the page takes 5 million to 4.2 million, which would need a further factor of 84 percent. A reader following only the text does not reach 4.2 million, so does not reach the INR 10,000 fixed cost per user, so does not reach the INR 15,000 floor. We publish both readings and leave the firm's own exhibit to settle it.

Six moves in a product launch framework, and the first two

The arithmetic above has an order, and getting the order wrong costs more than getting a multiplication wrong. Six moves. The sequence is ours, built from the case worked above and from the one criterion Oliver Wyman prints ahead of the arithmetic on its own pricing case.

1. Ask for the cannibalization rate before anything else. The firm's published criterion puts the ability to ask for key information ahead of the ability to calculate. In a launch case that information request is this one, and it is worth asking even when the answer is that nobody knows.
2. Ask what the existing line earns per unit. The displaced contribution needs two inputs and the rate is only one. Without the 38 cents, the 35 percent is a number you cannot spend.
3. Write the objective function, then the branches. Gross contribution, less displaced contribution, less fixed cost. Four branches hung under an equation beat five branches hung under a heading.
4. Compute gross contribution and stop. 40 million at 55 cents is 22 million dollars. Say the figure, say that it is gross, and say what has to come off it. Announcing the subtraction before performing it is what separates this from a rescue at the end.
5. Subtract, then solve for the rate that flips the answer. 52.6 percent, in this case. A point estimate invites an argument. A threshold hands the client something to go and measure.
6. Close on the thinness. 22 million dollars became 2.68 million. That ratio is the recommendation, and leaving it unsaid is how a correct calculation produces a wrong sentence.

Moves one and two belong first because they are the only two that need somebody else in the room. Everything after move three you can do alone, which means it can also wait. A candidate who computes for six minutes and then asks for the rate has spent the six minutes on an answer that is about to change.

The rubric lines a launch answer leans on

The weights that follow belong to us. They are read out of our own grading code in src/lib/grading, and no consulting firm publishes anything comparable. Why each line carries the weight it does is set out on how an answer is assessed.