How to sanity check assumptions in estimation interview questions
When answering estimation questions, always check your assumptions for realism and internal consistency. Cross-reference with known facts, break down big numbers, and state your reasoning out loud. This prevents wild guesses and helps you spot calculation errors or overlooked details before committing to an answer.
PMMockr.comWhy unchecked assumptions lead candidates astray
A common misstep in estimation interviews is confidently relying on an assumption that turns out to be wildly off-base. For example, estimating the annual market for dog food in the US by assuming every household owns a dog leads to a massive overstatement. This happens when candidates do not pause to question if each input makes sense, fits together, or matches what is known about the world.
Sanity checking is not just about catching absurd numbers—it's about actively looking for disconnects, stacking assumptions together, and asking, 'Does this total feel plausible?' Interviewers expect you to verbalize this process, not just to avoid mistakes, but to show structured thinking and intellectual humility.
Start with explicit, checkable assumptions
Begin any estimation by stating each input as a clear, testable statement. For example, if estimating the number of new coffee shops in a city per year, you might assume the city's population, average coffee shop size, and the rate at which new shops open.
Writing down these numbers makes it easy to check their origins and adjust if needed. This transparency also invites the interviewer to correct any that seem unreasonable, which is preferable to building a calculation on a hidden error.
- Assume New York City population: 8 million (explicit, checkable)
- Assume one coffee shop per 4000 residents (clear, plausible ratio)
Break down big numbers to spot inconsistencies
Large or complex numbers often hide mistakes. If your estimate for annual umbrella sales in the US comes out to 1.5 billion units, pause and break it down: that's over four umbrellas per person, per year. Does that seem reasonable?
Use this breakdown to test for reasonableness. If the number feels off, revisit your inputs. Perhaps you overestimated the frequency of purchases, or forgot about product lifespan. This step catches errors that are not obvious from looking at the final figure alone.
Cross-reference with familiar anchors
Whenever possible, compare your assumptions to numbers you know from daily life or public knowledge. For instance, if estimating the number of pizzas sold in the US, check against how many pizzas a family might order per month. Or, if you assume there are 300 million cars in the US, compare that to the US population (about 330 million). This sanity check uses context to avoid outlandish figures.
If you realize your estimate is far outside the range of such anchors, flag it and explain your reasoning, or adjust accordingly. This shows that your process is grounded and flexible.
Worked example: Estimating annual bicycle sales in San Francisco
Suppose you are asked: 'Estimate the number of bicycles sold in San Francisco each year.'
First, state your key assumptions: - San Francisco population: assume 900,000 - Percentage of people who own a bicycle: assume 20% (so 180,000 owners) - Average bike lifespan: assume 10 years - Assume most purchases are for replacement, not first-time buyers.
Now, calculate annual replacement demand: 180,000 bikes / 10 years = 18,000 bikes per year. Assume an additional 2,000 new buyers annually due to population inflow or shifting preferences, for a total of 20,000 bikes sold per year.
Sanity check: Is 20,000 plausible? If you assume 100 bike shops in the city, that's 200 bikes sold per shop per year, or about four per week. This seems reasonable for a city with a strong cycling culture.
Alternative: Suppose you had assumed 50% ownership (450,000 bikes) and a five-year lifespan, yielding 90,000 bikes sold per year. Breaking this down, each shop would need to sell 900 bikes per year, or around 17 per week—implausibly high given typical store capacity and demand.
Here, the initial lower assumption is better supported by the shop-level breakdown and public knowledge about bike usage. If you learned from a city report that only 10% of residents bike regularly, you would adjust your estimate downward. If you discovered a major cycling event led to a surge in purchases, you might revise upward. The key is that each assumption is open to adjustment, and your reasoning is transparent.
Assumption triangulation: Seek more than one justification
Strong assumptions are rarely based on a single source or guess. Instead, try to triangulate: support each number from two or more perspectives. For the bicycle example, you could justify ownership rates by referencing census data, local cycling infrastructure investment, or the number of bike racks in public areas.
When you explain an assumption, give at least one reason why it could be too high and one why it could be too low. This shows awareness of uncertainty and a willingness to revise as new data emerges.
Weak reasoning versus robust reasoning
A weak approach: 'I assume 50% of people in San Francisco own a bike because it’s a very hilly city.' This ignores the deterrent effect hills might have and lacks supporting context.
A stronger approach: 'I estimate 20% bike ownership. While San Francisco is known for cycling, its hills and compact size may limit mass adoption. City bike-share statistics and observed infrastructure can help anchor this estimate.'
The difference is that robust reasoning acknowledges both supporting and limiting factors, seeks multiple data points, and stays open to correction.
Validate with a practical experiment or cross-check
To validate your estimate, propose a concrete experiment or data check. For the bike sales example, you could check annual sales data from a few major bike retailers and multiply by the estimated market share. Alternatively, look up city transportation reports on cycling rates.
If you find that actual reported sales are much higher or lower, revisit your assumptions and adjust. This step demonstrates to interviewers that you are not rigidly attached to your first answer, and that you value evidence over guesswork.
What would change your recommendation?
State explicitly what evidence would alter your estimate. For example, 'If I learned that citywide bike sales were 35,000 per year according to a trade association, I would revisit my ownership and lifespan assumptions to align with this higher demand.'
This openness to change shows maturity and a data-driven mindset, key qualities for product managers making real-world decisions.
FAQ
How do I handle an assumption if I have no data or personal experience?
State the assumption clearly, explain your reasoning, and flag it as an area of high uncertainty. Where possible, anchor it to something you do know, and show how you would update if new data became available.
Is it ever acceptable to make a wild guess if I’m stuck?
If you must guess, do so transparently and explain it as a placeholder. Always break down the guess into smaller parts and look for ways to check if it is plausible or internally consistent.
What if my interviewer corrects one of my assumptions?
Thank them for the correction, incorporate the new information, and recalculate. This shows flexibility and willingness to learn, both important PM traits.