What the interviewer wants
A market-sizing question asks whether you can turn a vague market into a number someone could use. The score is on the tree, not on matching a hidden answer. You should:
- Clarify what you are sizing: revenue, units, or profit, and for which year and geography.
- Draw a driver tree with three or four levels, not a single multiply.
- State each assumption before you calculate.
- Check the result against a fact that would make it absurd.
The arithmetic itself is covered in the case math guide. This page is the shape of the estimate.
Two trees that cover most cases
Demand side. Population × share who buy × frequency × price. Use this for consumer markets. Segment before you multiply if the population is not one group. Saudi coffee drinkers in Riyadh are not the same as the national population.
Supply or account side. Number of accounts × penetration × annual contract value. Use this when the buyer is a company, a store, or a ministry. Counting people produces a nonsense market.
Pick one tree and stay in it. Mixing population and account counts in the same multiply is the usual way to be off by 10x.
Worked shape
A premium coffee chain sizing Saudi entry might say: urban adults in three cities, share who buy specialty coffee weekly, visits per week, ticket. That is a demand tree with a geography cut. Treating "Saudi Arabia" as one city, or using the European ticket with no currency or tax note, is the miss. The Saudi coffee case is that prompt, run as a full case rather than a standalone sizing.
Sanity checks that save you
- Per household or per store, does the spend look like a real basket?
- Is the market at least as large as the known players' revenue?
- If you change the weakest assumption by half, does the decision change? Say so.
State the number, the unit, and the assumption you trust least. That last sentence is what separates a sizing from a guess.