How Do You Answer a Market Sizing Interview Question?

Prompt and Applicable Context

Estimate one year of delivery-fee revenue for same-day prescription delivery in a fictional city. Without external data, how would you decompose, cross-check, and qualify the result?

This is a market-sizing question for product management, business analysis, strategy, and consulting roles. Public 2026 product-interview preparation materials still include market, revenue, and usage estimates, while Chinese product-interview material includes Fermi estimation as a case type. That evidence establishes current relevance without attributing the prompt to a particular company or claiming an unsupported interview frequency.

The prompt intentionally withholds real market data. The interviewer is not testing whether you remember how many pharmacies a city has. They want to see whether you define the problem before modeling it, preserve units, expose assumptions, test the order of magnitude through an independent path, and state which decisions the estimate can and cannot support.

Every population, household-size, prescription-frequency, eligibility, adoption, pharmacy-count, delivery-capacity, and price figure below is an interview practice assumption. None describes a real city or business.

What the Interviewer Evaluates

The first signal is metric control. “Market size” may mean users, orders, gross merchandise value, delivery-fee revenue, or gross profit. If the prompt asks for delivery-fee revenue, drug sales do not belong in the numerator, and revenue cannot be relabeled as profit. Time, geography, customer unit, and currency must also be fixed before calculation.

The second signal is structured decomposition. Each layer should be explainable: population becomes households, some households generate prescription demand, each generates a number of fills, some fills qualify for same-day delivery, some eligible fills adopt the service, and an order becomes revenue. Branches should be as mutually exclusive and collectively complete as the situation permits.

The third signal is uncertainty management. Assumptions are allowed when data is absent, but the answer should identify their evidence level, provide low, base, and high cases, and name the variable worth validating first. Extra decimal places do not create confidence. Explaining sensitivity to eligibility or adoption does.

Finally, the interviewer looks for cross-checking and a decision. A demand-side model asks how much customers might buy. A supply-side model asks how much the initial network could deliver. Similar results do not establish a real market, and widely different results should not be averaged. A strong answer traces disagreement to units, coverage, adoption, or capacity and designs the next investigation.

Questions to Clarify Before Answering

• Which metric is being estimated? This answer estimates annual delivery-fee revenue. It excludes drug GMV, membership revenue, refunds, subsidies, courier costs, payment fees, and profit. A GMV or profit question requires a different formula.
• What are the geographic and time boundaries? The exercise uses one fictional city and one year. City residents, metropolitan-area residents, and the serviceable area are different denominators. Daily, monthly, and annual figures cannot be mixed.
• Is the customer unit a person, prescription, or household? This model uses households as demand units and fill events as transaction opportunities. Population derives household count and must not be added again later.
• What qualifies for “same day”? Clarify whether the scope covers only otherwise eligible routine prescriptions or also cold-chain, controlled, emergency, and cross-region orders. This exercise isolates eligibility as an assumption rather than inventing regulatory facts.
• Is price a delivery fee or total order value? The exercise assumes an $8 fee per delivered order. A free-delivery threshold, subscription, or dynamic pricing model would require a paid-order share or realized revenue per delivery.
• Which decision will use the estimate? Early opportunity screening needs an order of magnitude and sensitivity. Investment, hiring, and launch require observed demand, competition, unit economics, regulation, and operating data.
• May external sources be used? If not, expose assumptions and ranges. If yes, use authoritative demographic and industry data for broad anchors, then interviews, surveys, historical orders, or experiments for product-specific adoption and willingness to pay.

30-Second Answer Framework

“I will estimate annual delivery-fee revenue, excluding drug GMV and costs. Demand starts with households and prescription frequency, then applies eligibility, adoption, and fee, with ranges on the two uncertain rates. I will cross-check supply using partner pharmacies, deliveries per day, and operating days. If the paths disagree, I will inspect units, coverage, and utilization rather than average them. The result can justify more research; launch still needs regulatory and unit-economic validation.”

In a full answer, write a unit-bearing formula before substituting figures. Every number should be reproducible through that formula, including the range and cross-check.

Step-by-Step Deep Answer

Step 1: Establish a scope, unit, and metric contract

Write one line first: fictional city × next year × eligible same-day prescription deliveries × delivery-fee revenue in U.S. dollars. Every later variable must map back to those four dimensions.

The revenue layer is the most common source of drift. Drug price belongs to merchandise value. The $8 delivery fee is service revenue. Courier payment, refunds, subsidy, and payment processing must be deducted before approaching contribution profit. Because the prompt asks only for fee revenue, the conclusion cannot say “the market earns $5.8 million in profit.”

Step 2: Build the demand-side formula and audit its units

Use this demand relationship:

Annual delivery-fee revenue = population ÷ people per household × share of households with prescription demand × fills per active household per year × same-day eligibility × adoption × fee per delivery

The explicit practice assumptions are:

• population: 5,000,000 people;
• household size: 2.5 people, producing 2,000,000 households;
• households with prescription demand during the year: 60%, producing 1,200,000 active households;
• fills per active household per year: 6, producing 7,200,000 fill events;
• same-day eligibility: 40%;
• adoption among eligible events: 25%;
• fee per delivery: $8.

Base-case orders are:

7,200,000 events × 40% × 25% = 720,000 deliveries

Annual delivery-fee revenue is:

720,000 deliveries × $8 per delivery = $5,760,000, rounded verbally to about $5.8 million.

Keeping units on every line shows that people first became households, households became fill events, and eligible adopted events became deliveries. Multiplying five million people directly by six household fills would fail the unit check.

Step 3: Express uncertainty as a range

Hold population, household size, frequency, and fee constant for this exercise. The weakest assumptions are eligibility and adoption. Set the low case to 30% eligibility and 10% adoption:

7,200,000 × 30% × 10% = 216,000 deliveries; 216,000 × $8 = $1,728,000, or about $1.7 million.

Set the high case to 50% eligibility and 40% adoption:

7,200,000 × 50% × 40% = 1,440,000 deliveries; 1,440,000 × $8 = $11,520,000, or about $11.5 million.

The practice conclusion is therefore about $1.7 million to $11.5 million in annual fee revenue, with a $5.8 million base case, while all other assumptions remain fixed. The wide range says the model is not ready to justify a heavy commitment. It does reveal which research could reduce decision risk.

Step 4: Test sensitivity instead of adding decorative assumptions

Hold everything else fixed and move adoption from 25% to 35%. That is a 40% relative increase. Base revenue moves proportionally from $5,760,000 to:

7,200,000 × 40% × 35% × $8 = $8,064,000.

Eligibility has the same linear property. Both are less supported than the population assumption, so they deserve validation first. An interview does not need a ten-level table for every input. Select two or three uncertain, high-impact variables and explain how they change the decision.

If the interviewer supplies a launch-cost threshold, solve backward for break-even adoption. Divide target revenue by $8 to get required orders, then divide by 7,200,000 fill events and eligibility. A threshold question often informs a decision better than polishing the point estimate.

Step 5: Cross-check the order of magnitude from supply

Create an independent capacity path. Assume 60 partner pharmacies in year one, an average of 35 completed deliveries per store per day, and 360 operating days:

60 stores × 35 deliveries per store per day × 360 days = 756,000 deliveries per year

At $8 each, that capacity corresponds to:

756,000 × $8 = $6,048,000, or about $6.0 million.

Supply exceeds the 720,000-delivery demand base by 36,000, or 5% relative to the demand base. This closeness was produced by practice assumptions. It is not evidence of real demand. Supply may confuse theoretical capacity with actual utilization, while demand may overstate adoption. Agreement only says the present arithmetic and scale have no obvious conflict.

If the paths differ threefold, align their contracts and inspect them line by line. Do they use the same service area? Does pharmacy count include only partners? Is 35 a peak rate or an annual average? Was adoption applied to all fills or eligible fills? Were operating days converted twice? Finding the disagreement is more useful than averaging the answers.

Step 6: Turn the estimate into a research plan

Secondary sources can calibrate broad denominators such as demographics, household structure, prescription volume, and pharmacy supply. The U.S. Small Business Administration's market-research guidance also distinguishes existing sources from direct research and calls out demand, market size, location, saturation, and pricing. Product-specific eligibility, adoption, and willingness to pay need primary evidence.

Prioritize the next steps by risk:

1. Confirm actual same-day eligibility with pharmacy and regulatory specialists to narrow that range.
2. Interview target users to separate wanting delivery from paying $8 for same-day delivery.
3. Test revealed behavior through a waitlist, concierge delivery, or small landing-page offer rather than relying only on stated interest.
4. Record completed deliveries, refusal reasons, and intraday variation across a small pharmacy cohort.
5. Add refunds, subsidies, fulfillment cost, and repeat usage to revenue so that unit economics can be evaluated.

The product value of an estimate is deciding which information to buy first. The estimate cannot prove a market or replace regulatory review, competitive analysis, or a real pilot.

Step 7: Put a decision boundary around the conclusion

A complete conclusion includes the point, range, sensitive variables, cross-check, and use: “Under the practice assumptions, annual delivery-fee revenue is about $5.8 million in the base case and $1.7 million to $11.5 million across low and high cases. The supply model implies about $6.0 million of capacity. Eligibility and adoption dominate uncertainty, so the next stage should validate them. The result supports a low-cost research decision, not a launch or investment level.”

This phrasing remains actionable without presenting model output as observation. As real evidence arrives, replace assumptions one by one and preserve model versions and variance explanations instead of changing only the final number.

High-Quality Sample Answer

“I will first define the metric: one year of realized same-day prescription delivery-fee revenue in a fictional city, in U.S. dollars. It excludes drug GMV, subscriptions, refunds, subsidies, and fulfillment cost. I will estimate demand and then cross-check supply. Every figure that follows is an interview assumption.

Assume five million residents and 2.5 people per household, which gives two million households. If 60% have prescription demand during the year, there are 1.2 million active households. At six fills per active household, that is 7.2 million fill events. If 40% are eligible for same-day delivery and 25% adopt, annual delivery volume is 720,000. At $8 per delivery, base fee revenue is $5.76 million, or about $5.8 million.

I would not report only that point. At 30% eligibility and 10% adoption, the low case is 216,000 deliveries and $1.728 million, or about $1.7 million. At 50% eligibility and 40% adoption, the high case is 1.44 million deliveries and $11.52 million. The practice range is therefore about $1.7 million to $11.5 million.

For a supply check, suppose year one has 60 partner pharmacies, each completing an average of 35 deliveries per day over 360 days. That is 756,000 deliveries and $6.048 million, or about $6.0 million. It is 5% above the demand base. The two practice models are compatible in scale, but that does not establish demand or prove full capacity utilization.

Eligibility and adoption are the sensitive assumptions. For example, moving adoption from 25% to 35%, with everything else fixed, moves revenue from $5.76 million to $8.064 million. I would next validate eligibility with pharmacy and regulatory specialists, then test adoption and willingness to pay $8 through user research and a limited paid pilot while measuring real fulfillment cost.

My conclusion is a $5.8 million base case and a $1.7 million to $11.5 million range under the stated assumptions. That is enough to decide whether a low-cost validation stage is worthwhile. It cannot by itself justify launch, hiring, or investment.”

The point is not to memorize five million or eight dollars. The interviewer should be able to replace any assumption and reproduce the output. If they provide real data, substitute it immediately while preserving the metric, units, and both model paths.

Common Mistakes

• Calculating before defining “market size” → Users, transactions, GMV, revenue, and profit become interchangeable → Write geography, time, customer unit, metric, and currency first.
• Switching between people and households → Demand is double-counted and formula units break → Label every step and state when people become households, events, and orders.
• Using decimals as evidence → Unsupported inputs remain unsupported at three decimal places → Round sensibly when speaking while preserving reproducible arithmetic.
• Reporting one point estimate → Uncertainty and failure boundaries remain hidden → Give low, base, and high cases for high-impact unknowns.
• Listing ranges without sensitivity → A wide interval does not guide research → Identify the variables that can move the outcome and lack evidence.
• Averaging demand and supply estimates → Averaging does not remove either model's bias → Trace the difference to coverage, units, adoption, or capacity utilization.
• Calling capacity demand → The ability to deliver 756,000 orders does not mean customers will buy them → Keep demand, capacity, and realized transactions separate.
• Calling delivery revenue GMV or profit → The metric layer changes and the product conclusion fails → Name inclusions and exclusions; build a separate cost model for profit.
• Recommending launch from the estimate → Regulation, competition, retention, and unit economics remain unknown → Limit the conclusion to the next research or pilot decision.

Follow-Up Questions and Responses

Follow-up 1: The 25% adoption assumption has no evidence. Why use it?

Call it an unvalidated assumption and do not defend the number. Show outcomes at 10%, 25%, and 40%, or work backward from the interviewer's target revenue to required adoption. The estimate should expose decision dependencies, not pretend adoption is known.

Follow-up 2: What changes if the service uses a monthly subscription instead of a per-order fee?

Revenue becomes eligible households times subscription adoption times monthly price times 12, adjusted for average subscribed months or churn. Order volume still checks fulfillment capacity, but it can no longer be multiplied by $8 as revenue. Free-delivery usage intensity also affects unit economics.

Follow-up 3: What if regulatory limits reduce eligibility from 40% to 20%?

With every other assumption fixed, orders and revenue halve: 7.2 million fills times 20% times 25% gives 360,000 orders, and multiplying by $8 gives $2.88 million. Then reassess excess supply and whether a partner network remains economical at the lower scale.

Follow-up 4: What if demand and supply differ by three times?

Do not average them. Align geography, period, and order definition. Check whether supply uses peak capacity or average utilization, whether demand applies adoption only to eligible events, and whether pharmacy coverage is complete. If the gap remains unexplained, preserve it as a risk range and design the cheapest test that distinguishes the models.

Follow-up 5: How would you update the model if real order data became available?

Calculate observed fill frequency, eligibility, adoption, paid amount, repeat usage, and refunds by user or household cohort. Separate new and existing users and service areas. Replace corresponding assumptions while retaining the old version and explaining variance. Check whether promotional periods or a limited pharmacy sample create selection bias before citywide extrapolation.

Follow-up 6: How should this result affect the product roadmap?

First ask whether the conservative case can still cover validation and fulfillment costs. If adoption drives the answer, prioritize willingness-to-pay and revealed-conversion experiments. If eligibility drives it, finish regulatory and supply research first. Broader building begins only after the critical range narrows and both user value and unit economics hold.