Approach Decision Tree: Your First 30 Seconds in a Guesstimate
The previous move, sanity-checking against something you know, separates a good guesstimate from a great one. But before you can sanity-check, you need the right approach in the first place. The first 30 seconds of a guesstimate are where you clarify scope, choose top-down or bottom-up, sketch the tree, and signal that you will cross-check before committing to a final number.
- Top-down starts from a macro base such as population, total market, or total users, then narrows through percentage filters.
- Bottom-up starts from one unit such as one customer, one outlet, one transaction, or one company, then multiplies up.
- Your first 30 seconds should produce a clear scope, a chosen approach, a visible estimate tree, and a promised cross-check.
- Use top-down when the problem has a reliable macro anchor, such as annual 2W sales of 1.9 cr or WhatsApp MAU India of 55 cr.
- Use bottom-up when supply, capacity, transaction volume, or named players give a cleaner build, such as Ola, TVS, Bajaj, and Ather in EV 2W sizing.
- A strong answer presents a range, not just a point estimate, and invites the interviewer to challenge assumptions.
The First 30 Seconds: The Whole Decision Tree
Before the arithmetic starts, the interviewer should hear your operating plan. This is not a speech - it is a compact decision tree that turns an ambiguous market-sizing prompt into a structured calculation.
Why the Opening Matters
A guesstimate is a structured estimate under uncertainty, not a memory test. The source glossary defines Fermi Estimation as rapid order-of-magnitude estimation with little data, named after Enrico Fermi. That means the interviewer is evaluating how you break down the problem, not whether you happen to know the exact market number.
The opening matters because it prevents three common failures: solving the wrong scope, mixing incompatible units, and discovering too late that your approach has no cross-check. The Board Infinity 8-step universal recipe starts with: clarify scope, choose approach, sketch the tree, state assumptions, calculate, sanity-check, present a range, and invite probe. The Approach Decision Tree is the compressed version of the first three steps plus a promised sanity-check.
Top-Down vs Bottom-Up: The Core Choice
Top-down means starting from a macro figure and narrowing it using percentage filters. Example: start with a total addressable user base, then apply penetration, activity, adoption, and frequency. Bottom-up means starting from a unit and multiplying up. Example: start with one outlet, one vehicle, one business, one transaction, or one company target, then aggregate.
Neither approach is automatically better. In many interviews, the stronger answer uses one as the main calculation and the other as the cross-check. Your decision should depend on which side has cleaner anchors in the prompt or in your memorised anchor set.
The Four-Move Opening Script
In an interview, the Approach Decision Tree should sound natural and concise. You are not asking for permission to think; you are showing the interviewer how you will think.
How to Choose the Right Branch
The cleanest way to choose is to ask: where is the most reliable anchor? An anchor is a starting number that gives the estimate stability. For example, the source gives India annual 2W sales as 1.9 cr, active 2W fleet as ~20 cr, WhatsApp MAU India as 55 cr, and Flipkart DAU as ~5 cr in the feature-impact example.
If the question is about a market or population, top-down is usually efficient. If the question is about capacity, operating throughput, or named-company contribution, bottom-up often becomes more defensible. If the question is product-oriented, the source warns that product managers should decompose the funnel rather than size a feature as only DAU × ₹.
Ola: A Cross-Checked EV 2W Estimate
The electric vehicle 2W example shows the full Approach Decision Tree because it has a macro demand base, a penetration assumption, a revenue conversion, and a named-player cross-check. EV means electric vehicle, FY means financial year, and 2W means two-wheeler.
The shallow answer is just 2.1 crore × 18%. The complete answer explains why top-down is the primary branch, why named companies are the cross-check, and why the final answer is a range.
Product Guesstimates Need a Funnel, Not Just a Base
Product management guesstimates use the same decision tree, but the building blocks shift. DAU means Daily Active Users, MAU means Monthly Active Users, and the DAU/MAU ratio measures stickiness, or how often monthly users return. The source notes that PM interviews are not only testing market-sizing rigour; they test whether you can size a feature, user segment, or revenue lever inside a known product.
GMV means Gross Merchandise Value, or the total value of transactions on a marketplace, not the marketplace revenue. This distinction matters in the Flipkart example because the source reports incremental GMV, not profit or net revenue.
To clarify, I am sizing X in India for a defined time period, in the requested unit. I will use top-down or bottom-up because that anchor is cleaner, my tree is base × filters × frequency × value, and I will cross-check using the opposite lens before giving a range.
Structuring The Approach Decision Tree Interview Answer
"Estimate the FY28 revenue opportunity for electric two-wheelers in India."
The fastest way candidates lose structure is by calculating before naming the tree. Say the approach and cross-check first; the arithmetic should feel like execution, not exploration.
Conclusion
The Approach Decision Tree is the discipline of making the first 30 seconds count: clarify the scope, choose the cleaner branch, sketch the equation, and promise a cross-check. In strong guesstimates, the final number matters, but the visible decision process is what makes the answer interview-ready.
The most frequent error is starting with numbers before choosing the approach. It costs points because the interviewer cannot tell whether your estimate is top-down, bottom-up, or just a chain of guesses, and it makes later cross-checks look like damage control instead of planned validation.