How to avoid excessive detail when market sizing?

Great question! Finding the right level of detail in market sizing is all about balancing structure, realism, and time pressure.

1. Start simple, then layer if needed

Always begin with a clear, top-down estimate. This shows you can think in a structured way. Only go deeper if prompted or if the case context suggests it’s worth refining.

2. Match detail to context

Ask yourself:

• Is this a quick estimate or will it guide a real decision?
• Would segmentation add real value or just slow me down?

For broad questions (like global phone sales), a high-level approach is usually sufficient, unless the interviewer asks for more precision.

3. Communicate your plan upfront

Say something like: "I’ll start with a top-level estimate using global averages, then segment by region or age group if needed."
This shows awareness without overcomplicating.

4. Use segmentation when it’s meaningful

Only add layers (e.g., different replacement cycles or prices) if the segments behave significantly differently.

Example: Estimating global phone sales revenue

Basic approach:

• Global population = ~8B
• Phone ownership = ~70% → 5.6B users
• Replacement every 3 years → annual sales = 5.6B / 3 ≈ 1.87B phones
• Average price = $500 → Revenue ≈ 1.87B × $500 = $935B

Optional refinement (if time or prompted):
Segment by region:

• Developed markets (1.5B users, $700 avg, 2-year cycle)
• Developing markets (4.1B users, $300 avg, 4-year cycle)
  Apply those separately, then sum

Bottom line:

Start clean and top-down, and only go granular when it adds value or helps clarify major differences. That’s what interviewers are really looking for.

That's a great question, unfortunately without a straightforward answer. 

One way I typically recommend candidates to look at it is as a set of filter.

Meaning, let's say you start top down from the population. You then look at the number of relevant people within the population due to age (filter 1), then at a subsegment that is relevant for a specific reason (filter 2)...

Now, if you only have 2-3 filters, the approach is a bit too light. 

If you have 15 filters, the interviewer will either go to sleep or stop you. 

So ideally, aim for somewhere in the middle: 5-10 filters.

Needless to say, this is only directionally correct. It all depends on the market sizing question.

At the end of the day, the whole exercise is meant to be a back-and-forth, a conversation with the interviewer and you are meant to jointly agree on an approach. Meaning, the set of filters and how many are necessary should develop naturally. 

Best,
Cristian

Hey there,

Aim for a level of detail where you can support your assumptions based on reasonable arguments (not a complete guess) and where you don't need to know a lot of information from memory (only basics, e.g., total population). 

E.g., how do you know there is 70% phone penetration rate globally? is this a guess or do you have any arguments? is this for all phones or only smartphones? If you focus on a geography you are familiar with, let's say you grew up in US, you can say this about smartphones:

• 0-20 years old - Penetration rate is maybe 50-60% because kids under 9-10 probably don't have a phone and you'd expect most 10-20 to have one
• 20-40 years old - You'd expect 95-100% of this population to have a phone (developed country, 20-40 most active study/work life, easy to obtain phones in US due to telecom market structure)
• 40-60 years old - You'd expect maybe 85-95% (while still active study/work life, maybe some of the older people in this category still use feature phones and not smartphones)
• 60+ - You'd expect many people in this group started adopting the technology, but less than 40-60, so maybe 75-85%

As you can see, adding the geography (US) and age groups helps you bring it closer to something familiar and offer reasonable assumptions based on your observations. It also helps refine the assumptions in comparison to each other (e.g., behavior for 0-20 years old vs 20-40 vs 40-60 vs 60+). If you need to calculate the global number, you can think what share of global population would have a similar behavior to US and add a discount for the rest (to keep it simple).

Regarding price and replacement rate - whether to do it by group or overall depends on the behavior related to the market you are trying to size. Is there a different behavior regarding price in your groups? Probably not much (maybe younger people prefer the trendier phones). Is there a big difference in prices for smartphones? Yes, Apple is generally more expensive than other brands and has high market share in US, so when you come up with a price consider this. Is there a difference in replacement rate between age groups / brands? Probably not. 

Best,

Maria

Hi,

Try to start simple with the market sizing and only go one level deeper when it makes a major difference for the outcome. As a rule of thumb, 2-3 levels should be sufficient to do a proper market sizing. 

In your example: 

- level 1: price and volume (incl. replacement rate)

- level 2: segment per development level (price difference) and per age (volume difference)

Feel free to reach out via chat.

Mattijs

Your "detailed" approach seems to me to have the right level of complexity. I only question whether you need different replacement cycles per segment - that may be the single piece of "overcomplication"