What estimation tests
Estimation evaluates whether the candidate can define an unknown quantity, build a logical model, choose reasonable assumptions, calculate, and check the result. The exact answer is often less important than the structure.
Define the quantity
Clarify geography, population, product, channel, time period, units, and whether the answer is volume, revenue, capacity, or value. “How many coffees?” is incomplete without location and period.
Top-down method
Start with a broad population or market and narrow by relevant percentages. Example: population × share of adults × share who drink coffee × purchases per week.
Bottom-up method
Start with supply units and throughput. Example: number of coffee shops × transactions per shop per day × operating days.
Demand and supply reconciliation
Using two methods can expose an assumption error. If demand implies twice the volume that supply capacity can serve, revisit store count, throughput, or consumption.
Assumptions
Use rounded numbers and explain their basis. Distinguish known facts from assumptions. Keep units attached to calculations.
Range and sensitivity
Identify the most sensitive assumption and provide a reasonable range. False precision doesn't improve the estimate.
Capacity cases
For hospitals, factories, airports, or service operations, use units × capacity × utilization × time. Consider bottlenecks and whether theoretical capacity is usable.
Revenue and value
After estimating volume, multiply by price or value per unit with consistent definitions. Gross merchandise value, revenue, and profit are different outputs.
Reasonableness check
Compare the result with household behavior, known company scale, physical capacity, or another independent reference. If the result implies an impossible frequency or market share, revise the model.
Interpretation
Explain what data would be collected first in real work and how the result affects the decision. Estimation is a starting model, not a substitute for evidence.
Worked demand-side example
To estimate annual restaurant meals purchased in a city of one million people, assume 800,000 residents eat restaurant meals, averaging 1.5 meals per week. That produces about 62 million resident meals annually. Add commuters and visitors, then separate dine-in, takeout, and delivery if the decision requires channel size.
Worked supply-side cross-check
Assume 2,000 restaurants, 90 transactions per day, and 350 operating days. That gives 63 million transactions. The agreement with the demand estimate suggests the scale is plausible, while the underlying assumptions still need evidence.
Translating estimation into a decision
If the client is considering a new delivery service, total meals aren't the addressable market. The analysis must narrow to cuisines, customer segments, delivery adoption, average order value, commission, and serviceable geography. Market sizing becomes useful only when definitions match the business model.