Estimation: make assumptions visible and the math defensible
Build estimates from definitions, equations, units, explicit assumptions and independent plausibility checks. Finish with a range and a decision implication rather than a precise-looking unsupported number.
Original PMMockr learning guide · Examples are illustrative, not company interview transcripts.
Before you start
Practice multiplication and unit conversion; exact market knowledge is not required for a clearly labelled exercise.
What you will learn
- Define the unit, geography and period.
- Compute an auditable estimate and sensitivity range.
- Use an independent capacity check without averaging incompatible quantities.
Read the explanation, attempt the problem yourself, then compare the worked answer.
Estimate completed weekday trips for a fictional city bike service.
Original teaching scenario. Any figures are illustrative assumptions.
Scope and assumptions
- Illustrative demand: 200,000 eligible residents, 10% daily riders and 1.5 trips per rider per weekday.
- Illustrative capacity: 5,000 usable bikes and six trips per bike per weekday.
How to tackle this question
- Define the geography, day and completed unit. Write the equation and units before supplying any assumptions.
- Calculate step by step with rounded hypothetical inputs. Vary the assumption that drives the widest uncertainty.
- Check capacity independently and reconcile any mismatch. Summarize a range and the input you would research first.
Try a short outline and spoken answer before comparing.
Reveal model solution and reasoning
Demand estimate
200,000 people × 0.10 riders/person × 1.5 trips/rider/day gives 30,000 trips per weekday. These are exercise assumptions, not researched city statistics.
Capacity cross-check
5,000 bikes × 6 trips/bike/day also gives 30,000 trips/day. Agreement is a plausibility check, not validation: both sides depend on assumptions and may fail at peak times.
Range and decision
Keeping population fixed, 8% adoption at 1.2 trips gives 19,200 demand trips/day; 12% at 1.8 gives 43,200. With the stated capacity, realized trips would be bounded near 30,000 unless utilization or fleet changes. I would research peak availability and achievable trips per bike before sizing a launch.
Interviewer follow-up
Why not average demand and supply when they disagree?
They constrain different things. Completed trips cannot exceed either feasible supply or demand in the stated period. I would diagnose the mismatch and model peak bottlenecks rather than averaging them.
Check your reasoning
- Do all factors produce the requested unit?
- Are assumptions distinct from observed facts?
- Does my range respect operational capacity and peak demand?
Try a changed constraint
Repeat with a different daily adoption assumption and identify which additional measurement would most narrow the range.
Go deeper
Do not call a scenario range a confidence interval without a statistical model. Vary uncertain assumptions consistently and name dependencies.
More detailed segmentation is useful only when it changes usage or capacity assumptions enough to improve the decision.
Start with the vocabulary
Clarification
A question that resolves an ambiguity which could change your recommendation. Clarify the goal, audience, scope, constraints and time horizon; do not ask a long list of questions just to follow a template.
“Are we improving the experience for first-time riders or increasing repeat rides?” These goals could lead to different products.
Common mistake: Asking about button colors before knowing which user problem matters.
User segment
A group whose needs, behavior or constraints differ enough to require a different product decision. Compare segments, choose one and explain why it is worth prioritizing now.
In a navigation app, an unfamiliar-city commuter may need confidence about transfers; a familiar-route commuter may need disruption alerts.
Common mistake: Treating everyone aged 18–35 as having the same need.
Sensitivity and capacity check
Test how uncertain inputs change an estimate, and whether the result fits operational limits. A range should expose important uncertainty.
Daily bike trips depend on both willing riders and feasible trips per usable bike; compare the two constraints.
Common mistake: Averaging demand and capacity estimates or reporting a scenario range as a statistical confidence interval.
Build your answer, layer by layer
This is a learning sequence, not a mandatory order for presenting every answer. Adapt to the actual question.
Define the quantity
State exactly what you are estimating, including geography, population, unit and time period.
Apply it: Estimate completed rides per weekday in one city, not all ride requests worldwide.
Go deeper: Clarify whether cancellations and repeat users count.
Choose an approach
Pick top-down demand, bottom-up capacity or a hybrid that fits the information available.
Apply it: Estimate riders × trips per rider, then check against driver capacity.
Go deeper: Explain why the alternative method is a useful cross-check.
Write the equation
Define each factor before assigning numbers. Make sure multiplying the units produces the requested unit.
Apply it: People × eligible share × active share × rides per active person per day.
Go deeper: Spot where correlated factors could double count demand.
Segment the population
Separate groups with different usage patterns instead of assuming everyone behaves identically.
Apply it: Commuters, occasional riders and visitors can have different trip frequencies.
Go deeper: Balance useful detail against the time available.
State assumptions
Use round numbers and explain why each is reasonable. Label estimated values instead of presenting them as researched facts.
Apply it: “For this exercise, I will assume one in five eligible residents rides on a weekday.”
Go deeper: Identify which assumption has the weakest support.
Calculate with units
Show arithmetic in manageable stages and preserve units and powers of ten.
Apply it: Convert thousands of riders into daily trips before multiplying by fare.
Go deeper: Correct an arithmetic slip directly without rebuilding the whole answer.
Check capacity and plausibility
Compare the estimate with physical or behavioral limits and another method.
Apply it: Would the implied trips per driver require more working hours than a day contains?
Go deeper: Explain a disagreement between demand and capacity estimates.
Give a range
Vary uncertain assumptions to produce plausible low and high scenarios. A range is not a statistical confidence interval unless you have a model that supports one.
Apply it: Show lower and higher adoption scenarios around the base case.
Go deeper: Keep scenario assumptions internally consistent.
Find the sensitivity driver
Identify which uncertain input has the largest effect on the result.
Apply it: A small change in rides per active user may matter more than a minor population adjustment.
Go deeper: Prioritize the research that would narrow uncertainty most.
Explain the business implication
State the estimate, range, major assumption and how the answer would inform a decision.
Apply it: Use the estimate to size a pilot rather than treating it as a precise revenue forecast.
Go deeper: Explain what additional information is needed before committing investment.
Write the equation before choosing numbers
For hypothetical daily meal orders, use residents × ordering share × orders per ordering resident per day. With deliberately illustrative inputs of 10 million × 20% × 0.2, the result is 400,000 orders/day. Keep people distinct from accounts and completed orders distinct from attempts. If the question asks revenue, add realized value per order and define whether taxes, tips and payouts are included.
Stress the uncertain input and reconcile limits
Vary ordering share between 10% and 30%, keeping the other exercise assumptions fixed: demand spans 200,000–600,000 orders/day. Then estimate restaurant or courier throughput with its own units. If capacity supports only 300,000, realized orders cannot equal the higher unconstrained demand without a capacity change. The mismatch is useful information; averaging demand and supply would conceal it. State which input is most worth researching and finish with a rounded range suited to the decision.
Connect the reading to this case
An estimate should reveal its reasoning and survive a sanity check. Here, make each factor auditable before refining a small adjustment.
Julia Winn: Analytical interview mistakesEstimate completed food-delivery orders per day in Bengaluru.
Attempt this second problem before comparing. Behavioral answers must use a true experience.
Compare with PMMockr’s worked solution
Original public teaching answer; hypothetical figures and fictional experience are not results to claim as your own.
I would clarify the geography, day and unit: completed app-based meal orders in Bengaluru on an ordinary weekday, excluding groceries and cancelled orders. I would separate resident, commuter and visitor demand to avoid counting people twice.
Use a resident model: residents multiplied by the share who order, multiplied by orders per ordering resident per day. As deliberately hypothetical round inputs, 10 million residents × 20% ordering residents × 0.2 daily orders gives 400,000 orders per day. These inputs are arithmetic practice, not measured population or market data.
Vary ordering share from 10% to 30% while holding the other assumptions fixed: the estimate becomes 200,000 to 600,000. Then sanity-check with restaurant throughput and delivery capacity, and account separately for visitor demand. I would seek current official population and business data before using this model for an operating decision.
Pushback
What assumption would you research first?
The ordering rate drives a large part of the uncertainty. I would distinguish active ordering people from accounts and measure order frequency over an appropriate period before refining smaller adjustments.
Common mistake
There are millions of people, so roughly a million orders.
Why this answer is stronger
The stronger answer defines units, makes assumptions auditable, checks sensitivity and avoids presenting invented inputs as facts.
From a weak answer to a defensible one
Weak: My demand and supply estimates differ, so I will average them.
Stronger: They describe different limits. I will inspect utilization and turnover assumptions and bound realized orders by feasible capacity.
Prepare your own answer
Recalculate for a peak hour, making the conversion from daily demand explicit. Identify where double counting could enter.
Close the guide and answer aloud. After reviewing, identify one missing decision and retry on a different problem.
Watch one interview
Estimation question: Time spent at stop lights ↗
IGotAnOffer · Mock interview · YouTube
Watch with a task: Pause before the calculation and write units on every factor. Compare the candidate’s assumptions with your own.
Practice with a PMMockr worked answer
Open the practice problem and attempt it before comparing.
Compare after your attempt
Read the original worked answer, follow-up and feedback
Compare problem framing, assumptions, rejected options, evidence and response to pushback. Explain what evidence would settle a disagreement.
Further reading and perspectives
Official documentation explains product behavior; coaching and community accounts offer preparation perspectives. Interview examples do not confirm a company’s current hiring process.
- Thomas Varghese: PM interview playbook, estimation questions
Optional additional perspective on decomposing estimates.
Medium · practitioner guidance · Search excerpt reviewed; full article returned 403 - Jason Girouard: The PM Job Search, Practical Guide
Read one practitioner’s approach to preparation, reflection and answering before watching.
Medium · personal account · Public page - Julia Winn: Analytical interview mistakes
Connect the intended behavior change to a computable metric and explain the estimate. This historical account is not a current interview specification.
Medium · practitioner guidance · Public article reviewed
Turn the reading into practice
- Easy: outline your answer and explain every term you use.
- Medium: compare two options, defend one and name a downside.
- Hard: handle uncertainty, conflicting goals and a changed assumption.
Complete three questions at each level in your selected learning path where available. Review model solutions through the question’s solution action after meeting its access requirements.
Choose a learning path →Further reading
Continue with these related PMMockr articles. Examples in this guide are illustrative practice scenarios.