Specifically, an estimate of the annual number of electric scooters sold in a year. Does it make sense to segment the market by age clusters and then by potential customers — that is, people who live in cities and are either students or workers with commutes of 5–7 km from home? Once these people are identified, could you further segment them by their possible modes of transportation: car, bicycle, motorbike, public transportation and electric scooter? After estimating the number of electric scooters, divide by their lifespan to obtain an annual figure?
Back to overview
Be the first to answer!
Nobody has responded to this question yet.
hey there :)
For an interview this should clearly be top down rather than bottom up, because age clusters and commute mode trees quickly become too detailed and assumption heavy. A clean approach is to start from Italy’s population, narrow to urban adults, estimate the share using personal mobility for short distances, then apply an adoption rate for electric scooters based on comparable European markets and current penetration. You can then sanity check by dividing the installed base by average scooter lifespan to get annual sales. This keeps it structured, fast and interviewer friendly. Happy to discuss further if helpful.
best,
Alessa :)
There are many paths to reach this, top down/bottom up, or estimate usage first or production as yardstick
You’ll probably end up using steps that are easy to find numbers from, so that you can plug them into your approach
e.g. Start with the total number of road miles per year… exclude Miles from with four or more wheels… Exclude manual bicycle miles or recreational miles… exclude gas motorcycle miles… Exclude other things you can exclude out and you would have some kind of final estimate on electric scooter miles/year
Then you look at the growth rate and estimate how much growth happened each year, divide the growth (in miles) by average miles per scooter, and then you will have how many new scooters came into play
Many approaches will work, this is just the first one that came to mind
That is a very thorough thought process, which is excellent for brainstorming the drivers, but in a real interview setting, trying to structure a market sizing case this way will likely consume too much time and introduce too many assumptions that you cannot realistically support.
The biggest trap in your proposed method is the granularity and the confusion between calculating the total stock and the annual flow of sales. Trying to segment first by age clusters, then by commute type, and then by existing mode of transport (car vs. bus vs. bike) creates a model with far too many moving pieces. You'll spend 8 minutes setting up the structure only to have the interviewer stop you because it’s too complex to defend the penetration rates for each micro-segment.
The cleaner, more robust strategic approach is to focus on a classic demand-side driver model and ensure you separate replacement sales from new growth sales.
Start with the total relevant Italian population (urban density filter is smart). Filter by income and age suitability to get your total Serviceable Market. Apply a reasonable penetration rate (e.g., 4–8% ownership among this market) to calculate the total current stock of scooters in use. Your final annual sales calculation must then be the sum of two distinct figures: (1) Replacement Sales: (Total Stock / Average Scooter Lifespan) + (2) New/Growth Sales: (Total Stock * Annual Market Growth Rate). This structure is MECE (Mutually Exclusive, Collectively Exhaustive) and much easier to populate with defensible numbers.
Keep it simple and focus on the drivers, not the micro-segments. All the best!
That's one approach.
How well it goes depends on how effectively you generate assumptions and align them with the interviewer.
To develop your ability in this, as a drill, I suggest you always try to come up with three approaches for market sizing questions (just when you practice, not during the interview).
This way, with time, you'll develop a better understanding of the different patterns you can approach such problems with.
Best,
Cristian
