How to use top-down and bottom-up estimation in PM cases
When tackling estimation questions, start by picking top-down or bottom-up based on the data and constraints at hand. Top-down is fastest for big-picture sizing, while bottom-up is useful for granular planning. Clearly state your assumptions and sanity-check your math at each step.
PMMockr.comChoosing between top-down and bottom-up approaches
A common misstep in product interviews is defaulting to just one estimation approach—usually top-down—without considering whether it fits the question. The right choice depends on the level of detail required and the information available.
Top-down estimation works best when you have access to broad market or population data and are asked for a high-level figure, such as the total potential users of a product. Bottom-up is more appropriate when you can break a problem into discrete, countable units, such as estimating the number of rides a city’s taxi fleet completes daily. Stating your rationale for choosing a method shows you can adapt your estimation technique to the situation.
Understanding top-down estimation
Top-down estimation starts from a large, known figure and narrows it down through a series of logical filters. You might begin with the total population of a country, then apply percentages to estimate a target segment.
This approach is quick and helps set an upper bound, but it’s only as good as your segmentation logic and the realism of your assumptions. Overusing rough percentages without justifying them can weaken your answer. After finishing, sanity-check whether the result makes sense compared to what you know about similar products or markets.
Grasping bottom-up estimation
Bottom-up estimation works by identifying the smallest relevant unit and scaling up. For example, to estimate the number of daily coffees sold at a campus café, you might multiply the number of students by the average coffees each buys per day, then factor in operating days and seating capacity.
This method is useful when you have some on-the-ground knowledge or can make reasonable assumptions about behavior. It often produces a more defensible number, but the risk is missing a key driver or overcomplicating the calculation. Each step should be explained and linked to a plausible real-world activity.
Worked example: Estimating annual scooter rides in a mid-sized city
Suppose you are asked: "Estimate the annual number of rides for a shared electric scooter service in a city of 800,000 people."
First, clarify the goal: we want an order-of-magnitude estimate for business planning. Both top-down and bottom-up methods are possible, but let's proceed with both to illustrate their use.
**Top-down approach:** Assume 60% of the city is over 16 and eligible to ride (480,000). Let’s say 30% are interested in using scooters (144,000). If each interested person takes an average of 10 rides per year, we get 1,440,000 rides annually. These percentages are assumptions to be called out during the interview.
**Bottom-up approach:** Assume there are 2,000 scooters in the city (hypothetical). Each scooter is used 4 times per day on average. The service operates 365 days a year, so total annual rides = 2,000 scooters x 4 rides/day x 365 days = 2,920,000 rides.
Both methods yield different results. The top-down estimate is lower, possibly due to conservative participation rates. The bottom-up is higher, perhaps overestimating utilization. Discussing both helps triangulate a more credible figure and shows awareness of possible biases.
If new evidence showed scooters are concentrated in the city center and serve mostly tourists, I would adjust user segmentation or per-scooter utilization downward. I would validate by asking the company for anonymized ride data or checking public mobility reports.
User segmentation: making key assumptions transparent
Segmentation decisions drive both estimation methods. In the scooter example, the top-down approach required segmenting by age and interest, while the bottom-up approach implicitly assumed scooters were distributed and used evenly.
Being explicit about these assumptions is crucial. For instance, if the city has many students, the per-capita ride rate may be higher. If weather is poor for half the year, annual rides would drop. Invite correction from the interviewer by noting where a local factor could swing your estimate.
Prioritizing which method to use first
In interviews, time is limited. Decide quickly which method will get you to a reasonable answer fastest. If market-level numbers are easy to estimate, start top-down. If you have a clear sense of infrastructure or user activity, begin bottom-up. You can always sanity-check with the other method if time allows.
For the scooter case, if you know the fleet size or average usage, bottom-up gives a plausible answer fast. If not, estimating the addressable user base via top-down may be safer.
Trade-offs and alternative choices
Every estimation involves trade-offs between speed, realism, and risk of compounding errors. Top-down risks overestimating if you miss a key exclusion (e.g., minors, people without smartphones). Bottom-up can underestimate if you miss high-usage edge cases, like tourists or event surges.
A weak approach is to guess a number without showing your math or to copy a canned framework without adapting it. A stronger answer flexibly combines both methods, explains why you chose each, and checks the plausibility of the outcome.
Validation: how to check and adjust your estimate
After calculating, sense-check your answer. Does the number feel plausible compared to known figures? For the scooter example, compare your annual rides estimate to published numbers for similar cities, if available.
If your estimate is far higher or lower, revisit your key assumptions. Sometimes, a quick check with another method (e.g., per-scooter utilization versus total population) helps catch errors or reveal a more realistic range.
Timed practice exercise and self-review checklist
Set a timer for 10 minutes. Pick a product: "Estimate the number of reusable water bottles sold to college students in the US each year." Choose top-down or bottom-up. Clearly state your assumptions, walk through your math, and check your answer with the other method if possible.
Self-review checklist: - Did you clearly choose and justify a method? - Were your assumptions explicit and realistic? - Did you segment users thoughtfully? - Did you sanity-check your result? - Did you consider an alternative method or validation?
Aim to practice one guesstimate every few days, using the checklist to identify patterns in your reasoning and build confidence.
FAQ
How do I decide whether to use top-down or bottom-up estimation?
Choose top-down when you need a broad estimate and have access to big-picture data. Use bottom-up when you can break the problem into components you understand well. If unsure, start with the method that gets you closest to a realistic number quickly, and sanity-check with the other if time permits.
What if my top-down and bottom-up estimates are very different?
Large discrepancies usually point to an unrealistic assumption in one approach. Re-examine your key percentages, segment sizes, or unit utilization. Discuss these openly and suggest which figure you’d trust more and why, using real-world reasoning.
Can I use both methods in the same answer?
Yes, using both methods to triangulate your estimate demonstrates thoroughness. Briefly explain the logic behind each, compare results, and state which you’d use for planning. This shows you’re aware of uncertainty and can adapt your thinking.