Market Sizing and Estimation Questions

What is being tested

Not knowledge. The interviewer does not know the population of Belgium either, and the answer is not being checked against a source.

What is being assessed is whether you can take an unanswerable question, break it into things you can estimate, make assumptions a sensible person would accept, carry out the arithmetic correctly under mild pressure, and then notice if the answer is absurd. That sequence is the actual daily work of a junior consultant, which is why the exercise survives.

The two approaches

Top-down starts with a large population and filters it down. Bottom-up starts with a unit and multiplies up. Both are valid and they suit different questions.

The method, step by step

• Clarify the question. What exactly is being sized, in what geography, over what period, and in units or money? "The market for coffee" is three different numbers depending on the answers.
• State your approach before starting. "I will go top-down from the UK population, segment by age, then apply a usage rate and a price." The interviewer can then correct your route before you spend four minutes on it.
• Build the equation before you put numbers in. Write the chain of multiplications, then fill it.
• Choose round, defensible numbers. A UK population of 70 million rather than 68.3. Precision you do not have wastes time and buys nothing.
• Segment where behaviour genuinely differs and not otherwise. Splitting a population by age matters for coffee; splitting it by hair colour does not.
• Do the arithmetic out loud and keep track of your zeros, which is where almost every error occurs.
• Sanity-check the answer, then state it clearly with the one assumption that drives it most.

A worked example

Size the annual market for takeaway coffee in the United Kingdom.

Top-down. Population 70 million. Exclude the very young and the very old, leaving roughly 50 million plausible buyers. Assume a third are regular takeaway coffee buyers, so about 17 million. Assume they buy four cups a week, which is 200 a year, giving 3.4 billion cups. At an average of £3.50 a cup, that is roughly £12 billion a year.

Then check it. Twelve billion pounds across 70 million people is about £170 per person per year, or roughly £3.30 a week averaged across everyone including those who never buy. That is plausible. If the check had produced £2,000 per person, the error would be findable, and finding it is worth more marks than the original calculation.

The numbers worth knowing

A small stock of reference points removes the most common source of panic and lets you spend the time on the reasoning.

• Populations: UK roughly 70 million, US roughly 340 million, EU roughly 450 million, world roughly 8 billion.
• Age distribution in a developed country: usefully approximated as evenly spread across 0 to 80, so any ten-year band is about an eighth of the population.
• Households: UK around 28 million, US around 130 million. Average household size a little over two.
• A year: 52 weeks, 365 days, about 250 working days, about 2,000 working hours.
• Rough life spans: cars around 12 years, phones 3, laptops 5, fridges 12, mattresses 8. Replacement rate is one over the life span, which is the key move in most durable goods questions.

Where candidates lose marks

• Silent calculation. If the interviewer cannot hear the reasoning, there is nothing to mark and an arithmetic slip fails the whole answer.
• False precision. Using 68.3 million and then assuming a usage rate out of thin air is inconsistent, and it is slower.
• Over-segmentation. Five segments each needing three assumptions produces fifteen chances to be wrong and no more insight than two segments.
• Losing a factor of ten. Write the zeros down or work in millions throughout, and say the magnitude out loud at each step.
• No sanity check. An unchecked absurd answer is the worst outcome available, because it says you would present it to a client.
• Refusing to commit. "It depends on many factors" is not an answer. Give a number, then say what would change it.

Practising these

These improve faster than any other part of case preparation. Do one a day for three weeks, out loud, on a two-minute timer, and vary the type: consumer volumes, business markets, physical capacity, and the occasional strange one about a quantity nobody has ever counted.

Then check your reasoning rather than your answer. Where a real figure is available, the interesting question is not whether you were close but which assumption was wrong and by how much. That is the habit the exercise is trying to build.