Build the Decomposition Tree Before Any Arithmetic

In the previous lesson, the core habit was to state every assumption out loud. This lesson answers the next question: where do those assumptions sit before you start multiplying numbers? In strong guesstimate interviews, especially at product and business roles, interviewers reward a visible decomposition tree first because it reveals scope, drivers, assumptions, and sanity checks before arithmetic can distract from the structure.

Why the Tree Matters Before the Math

A guesstimate is not a mental-math contest. It is a structured reasoning exercise where the interviewer wants to see how you convert an ambiguous business question into an answerable model. Arithmetic is necessary, but it should come after the interviewer can see the shape of the problem.

The source compendium calls this the universal recipe: clarify scope, choose approach, sketch the tree, state assumptions with anchors, calculate with rounded numbers, sanity check, present a range, and invite a probe. The decomposition tree is the bridge between assumptions and arithmetic.

At a glance, a strong guesstimate answer moves through these stages before it reaches the final number.

Flipkart: The Full Framework in One Business

Flipkart is a useful example because product-management guesstimates often ask about feature impact, not just market size. In the source example, the question is how much a saved-for-later cart feature could move Gross Merchandise Value, or GMV, which means the total value of transactions on a marketplace.

A shallow answer would jump to “Flipkart users × ₹ value.” A complete answer shows the product funnel: who sees the feature, who adopts it, how behavior changes, and what each shifted action is worth.

How a Decomposition Tree Works

The tree works by forcing you to separate the question into layers. The first layer is the root: what exactly are you estimating? The second layer is the approach: top-down, bottom-up, funnel, or supply-side. The third layer contains measurable drivers. The fourth layer contains assumptions and checks.

Top-down means starting from a macro figure and narrowing through filters. Bottom-up means starting from one unit, such as one restaurant, one seller, one MW of capacity, or one vehicle, and multiplying upward. A funnel is a sequence of user actions, such as signup, onboarding, activation, retention, and monetization. A sanity check is a plausibility test against another benchmark.

The Product Management Version of the Tree

PM means Product Management. In PM interviews at Microsoft, Google, Flipkart, Meesho, Swiggy, and PhonePe, the source notes that guesstimates are not mainly testing market-sizing rigour. They test whether you can size a feature, user segment, or revenue lever within a known product.

That is why PM trees often use DAU and MAU. DAU means Daily Active Users, the number of unique users active in a day. MAU means Monthly Active Users, the number of unique users active in a month. The DAU over MAU ratio is a stickiness measure: how frequently monthly users come back on a daily basis.

ARPU means Average Revenue Per User, or the average revenue earned from each user over a defined period. In a marketplace or app feature estimate, the interviewer is usually listening for whether you decomposed the user base into a behavior funnel before converting behavior into rupees.

Worked Example: FY28 Electric Two-Wheeler Revenue

Consider the situation: estimate the FY28 electric two-wheeler, or EV-2W, market in India. EV means electric vehicle, and 2W means two-wheeler. The problem looks numerical, but the correct first move is to build a tree: total two-wheeler market, growth to FY28, EV penetration, units, average selling price, revenue, and sanity check.

The learning is that the final revenue number becomes credible because the interviewer can inspect each branch. If they disagree with EV penetration, average selling price, or player-level capacity, they can challenge one assumption without breaking the entire answer.

Choosing the Right Tree for the Question

Not every guesstimate deserves the same tree. A market-size question, a product funnel question, and a unit-economics question have different natural drivers. The mistake is to force every problem into population × penetration × price.

CPM means cost per mille, or cost per thousand ad impressions. KYC means Know Your Customer, the identity verification process used by fintech and financial products. These terms matter because using them correctly shows you understand the business system behind the estimate.

A Reusable Answer Template

Use this template whenever an interviewer gives you an ambiguous estimate. It slows you down just enough to avoid premature arithmetic and gives the interviewer a clean map of your thinking.

1. Clarify scope: Say the geography, time period, unit, and metric before choosing numbers.
2. Choose approach: Explain why top-down, bottom-up, funnel, or supply-side logic fits the question.
3. Sketch the tree: Say the root and branches out loud before multiplying.
4. State assumptions with anchors: Use rounded values and explain what each value represents.
5. Calculate cleanly: Use two significant figures where possible instead of over-precise math.
6. Sanity check: Validate through a second path, such as named players, per-capita logic, or known benchmark.
7. Give a range: Present a plausible band rather than pretending one point estimate is exact.
8. Invite a probe: Ask which assumption the interviewer would like to flex.

Structuring a Build the Decomposition Tree Before Any Arithmetic Interview Answer

Conclusion

A decomposition tree turns a vague guesstimate into a testable business model. If you clarify the root, map the drivers, state assumptions, calculate cleanly, and sanity check with a range, you show the interviewer the reasoning they are actually grading.