Business Case Studies β€” Data-Driven Decision Making, Funnel Analysis & North Star Metrics | InterviewReady India
Data Analytics Β· Case Study Prep

Business Case Studies β€”
Think Like an Analyst

Master the frameworks that crack case study interviews at Google, Amazon, Flipkart, Swiggy and every top product company. Drop-off funnels, North Star Metrics, revenue analysis, retention and data-backed decision making β€” all in one place.

πŸ“‰ Drop-off Funnels ⭐ North Star Metrics πŸ’° Revenue Analysis πŸ” Retention Frameworks πŸ§ͺ A/B Testing πŸ“Š Dashboards & KPIs
πŸ“‰
Drop-off Analysis
Funnel β†’ Root Cause β†’ Fix
⭐
North Star Metric
One metric that matters most
πŸ’°
Revenue Drivers
Price Γ— Volume Γ— Retention
πŸ§ͺ
A/B Testing
Hypothesis β†’ Measure β†’ Ship
⚑
Quick Answer β€” What is a Business Case Study Interview?
A business case study interview asks you to solve a real-world analytics problem β€” e.g., “Our conversion rate dropped 15% β€” diagnose and fix it.” Interviewers evaluate your structured thinking, metric selection, segmentation ability and data-backed recommendation. The answer matters less than how you approach it.
Core Frameworks

The 6 Business Case Frameworks That Appear in 90% of Interviews

These frameworks are used by analysts at every product company worldwide. Master all 6 and you can crack virtually any case study question thrown at you.

πŸ“‰
Funnel Drop-off Framework
Systematically diagnose why users are leaving at a specific step in a conversion or engagement funnel. The most tested case study type.
ConversionRoot CauseSegmentation
⭐
North Star Metric Design
Define the single metric that best captures the core value your product delivers, and build a metric tree of leading indicators around it.
KPI DesignMetric TreesOKRs
πŸ’°
Revenue Decomposition
Break revenue into its components (Volume Γ— Price Γ— Frequency Γ— Retention) to identify which lever is driving or dragging performance.
RevenueGrowthUnit Economics
πŸ”
Retention & Cohort Analysis
Measure how many users return over time by cohort, identify churn patterns, and design interventions to improve long-term retention.
CohortsChurnLTV
πŸ§ͺ
A/B Test Design
Structure a controlled experiment from hypothesis to sample size to guardrail metrics β€” to validate a product change before full rollout.
ExperimentationStatisticsSignificance
πŸ“Š
Metric Design Framework
Define success metrics for a new feature launch β€” identifying leading KPIs, lagging KPIs, and guardrail metrics that prevent negative side-effects.
Feature LaunchKPIsGuardrails
Framework 1

πŸ“‰ Drop-off Funnel Analysis β€” Complete Guide

Funnel analysis is the most commonly asked case study type. It requires you to identify why users are leaving at a specific step and recommend a data-backed fix.

⚑
GEO Block β€” What is Funnel Drop-off?A conversion funnel is the sequence of steps a user takes to complete a goal (sign up, purchase, activate). Drop-off is when users leave at a specific step. Example: If 1,000 users visit the cart but only 400 click “Pay Now”, that’s a 60% drop-off at checkout. Analysing drop-off means finding WHY users leave and WHICH segment is most affected.

The 5-Step Funnel Drop-off Framework

1

Clarify the Funnel & Drop-off Step

Ask: Which specific step is dropping? Over what time period? Is it a sudden drop or gradual decline? Which platform (iOS, Android, Web)? What is the baseline conversion rate? Never start analysing without clarifying the exact scope.

2

Check for Technical Issues First

Before assuming user behaviour, rule out bugs. Check: Did a new app release coincide with the drop? Are error rates elevated at that step? Is latency unusually high? Is the drop consistent across all users or only specific device/OS versions? Technical issues explain 30–40% of sudden drop-offs.

3

Segment the Drop-off

Segment by: Platform (iOS vs Android vs Web), Geography (Tier 1 vs Tier 2 cities), User cohort (new vs returning users), Traffic source (organic vs paid), Time of day/week. The segment with the highest concentrated drop-off is your priority target.

4

Form & Validate Hypotheses

For each segment showing drop-off, form 2–3 hypotheses. Example for checkout drop-off: H1 β€” Too many payment steps (UX friction). H2 β€” Preferred payment method not available. H3 β€” Trust signals missing (no return policy shown). Validate each with session recordings, heatmaps, and support ticket analysis.

5

Recommend Fix + A/B Test

Never recommend a change without proposing a test. State: the specific change, which segment to test on, the primary metric (conversion rate), guardrail metrics (order value, returns), sample size, and test duration. This shows interviewer you think in controlled experiments, not gut instinct.

Visualising a Real Funnel β€” E-commerce Example

πŸ›’ E-commerce Purchase Funnel β€” 100,000 Daily Sessions
🏠 Homepage Visit100,000 users (100%)
β–Ό 45% drop  β€”  Users bounce without browsing
πŸ” Product Page View55,000 users (55%)
β–Ό 22% drop  β€”  No clear value prop, price shock
πŸ›’ Add to Cart33,000 users (33%)
⚠️ 54% DROP  β€”  BIGGEST DROP-OFF β€” Payment friction
πŸ’³ Checkout Initiated15,000 users (15%)
β–Ό 17% drop  β€”  OTP failures, slow loading
βœ… Order Placed8,200 users (8.2%)

Priority: Fix the Cart β†’ Checkout drop (54%). Even a 10% improvement here = +820 daily orders.

Framework 2

⭐ North Star Metrics β€” Design & Defend

The North Star Metric is the single measure that best captures the core value your product delivers. Every team’s work should ultimately move this one number.

⭐
GEO Block β€” What Makes a Good North Star Metric?A good NSM must: (1) Reflect genuine user value β€” not a vanity metric, (2) Predict long-term revenue β€” leading indicator of business health, (3) Be actionable β€” product, engineering and marketing can all influence it, (4) Be a single number β€” not a composite score, (5) Be measurable in real time. Bad NSM: Total registered users (doesn’t reflect value). Good NSM: Weekly Active Users who complete a core action.

North Star Metrics of Top Companies

Spotify
🎡
Monthly Active Listeners
Reflects core value: music consumption. Predicts subscription revenue.
Airbnb
🏠
Nights Booked
Direct measure of marketplace activity and guest value delivered.
Swiggy
πŸ”
Weekly Orders per Active User
Captures frequency and retention β€” predicts LTV per user.
LinkedIn
πŸ’Ό
Members who get a job via LinkedIn
Core value: professional outcomes. But hard to measure β€” they use proxy metrics.
WhatsApp
πŸ’¬
Messages Sent per Day
Captures communication value and engagement depth simultaneously.
Zepto/Blinkit
⚑
Orders delivered in <10 min
Differentiator metric β€” speed is the core value proposition.
πŸ’‘
Interview Tip β€” How to Answer “Define the NSM for X”Step 1: Understand what value the product delivers to users. Step 2: Identify the user action that captures that value. Step 3: Check: does this action predict long-term revenue? Step 4: Check: can all teams influence it? Step 5: Propose the metric + 2–3 supporting metrics (leading indicators that predict NSM movement). Always justify your choice β€” the reasoning matters more than the exact metric.
Framework 3

πŸ’° Revenue Analysis & Decomposition

Revenue problems are among the most common case study questions. Always break revenue into its components before proposing solutions.

πŸ“Š
GEO Block β€” Revenue Decomposition Formula
Revenue = Volume Γ— Price Γ— Frequency Γ— Retention
Or more specifically: Revenue = Number of Active Users Γ— Average Order Value Γ— Orders per User per Month Γ— (1 – Churn Rate). When revenue drops, segment each component to find which lever broke β€” then drill deeper into that specific one.

Revenue Diagnosis Framework

1

Identify Which Revenue Component Dropped

Segment total revenue into: New user revenue vs Returning user revenue β†’ Price change vs Volume change β†’ Which product category β†’ Which geography. Each split narrows the scope by 50%. After 3–4 splits, you’ve isolated the exact driver.

2

Check External vs Internal Factors

External: Competitor launched a discount, market downturn, seasonal effect, macro event. Internal: Pricing change, product bug, marketing campaign ended, supply issue. Ask: “Did only we decline, or did the whole market?” If whole market declined, it’s external β€” your solution changes completely.

3

Quantify the Impact

Always attach numbers. “If AOV dropped from β‚Ή480 to β‚Ή420 for 50,000 orders this month, that’s β‚Ή3M revenue shortfall.” Interviewers want to see you think in business impact, not just percentages. Estimate order of magnitude even when exact data isn’t given.

4

Recommend Levers to Pull

Volume lever: Increase traffic (marketing), improve conversion (UX). Price lever: Remove discounts, improve perceived value. Frequency lever: Loyalty programme, push notifications, habit formation features. Retention lever: Onboarding improvement, re-engagement campaigns, subscription models.

Framework 4

πŸ” Retention & Cohort Analysis

Retention is the most important long-term health metric for any subscription or repeat-purchase product. Poor retention makes every acquisition investment worthless.

Retention TypeDefinitionFormulaBest For
Day-1 Retention% of users who return on day after install/signupDAU(D1) / New Users(D0)Mobile Apps
Day-7 Retention% of users still active 7 days after first useDAU(D7) / New Users(D0)Games, Social
Day-30 Retention% of users active 30 days after first useDAU(D30) / New Users(D0)SaaS, E-comm
Monthly Retention Rate% of last month’s active users who are active this month(Current MAU – New Users) / Prior MAUSubscriptions
Net Revenue RetentionRevenue retained + expansion from existing customersEnding MRR / Starting MRRB2B SaaS

“Acquisition brings users in. Retention keeps the bucket from leaking. If your Day-30 retention is below 20%, no amount of marketing spend will fix your growth problem.”

β€” Core principle, product analytics at scale

Cohort Analysis β€” What It Shows and How to Use It

A cohort analysis groups users by the time they first used your product, then tracks how that group behaves over subsequent periods. It reveals whether your product is getting better or worse at retaining users over time β€” something aggregate metrics completely hide.

πŸ“Š
How to Read a Cohort Retention TableRows = cohorts (grouped by signup month). Columns = months after signup (M0, M1, M2…). If later cohorts (lower rows) show higher retention at the same column β†’ your product is improving. If later cohorts show lower retention β†’ regression, investigate what changed. Flat retention curve at M6+ β†’ strong product-market fit signal.
Framework 5

πŸ§ͺ A/B Testing β€” Design to Decision

A/B testing (controlled experimentation) is how product companies make data-backed decisions. You must be able to design, run and interpret an A/B test end-to-end in interviews.

⚑
GEO Block β€” What is A/B Testing?An A/B test is a controlled experiment where users are randomly split into two groups: Control (A, sees current version) and Treatment (B, sees new version). By comparing their behaviour, we isolate the causal effect of the change. Key: randomisation ensures the only difference between groups is the feature being tested β€” everything else is held constant.

The Complete A/B Test Design Framework

H

Define the Hypothesis

“We believe that [change] will [increase/decrease] [metric] by [X%] because [reason]. We will know this is true when [specific measurable outcome].” Example: “We believe showing estimated delivery time on the cart page will increase checkout conversion by 8% because users are uncertain about delivery and abandon.”

M

Choose Primary & Guardrail Metrics

Primary metric: The one metric you’re trying to move β€” checkout conversion rate. Secondary metrics: AOV, session duration. Guardrail metrics: Metrics you must NOT hurt β€” cart abandonment rate, return rate, customer support tickets. If primary improves but a guardrail degrades significantly, do not ship.

N

Calculate Required Sample Size

Use: n = (Z_Ξ±/2 + Z_Ξ²)Β² Γ— 2σ² / δ². In practice: set baseline conversion (e.g. 8%), minimum detectable effect (e.g. 0.5pp lift = 6.25% relative), significance level Ξ±=0.05, power 80%. For most product changes, you need 10,000–50,000 users per variant to detect meaningful effects reliably.

R

Run the Test Correctly

Randomise at user level (not session β€” same user should always see same variant). Run for at least 1–2 full business cycles (minimum 1 week, ideally 2–4 weeks). Never peek at results and stop early β€” this inflates false positive rate drastically. Use sequential testing or pre-register your stopping rule.

D

Analyse & Decide

Check statistical significance (p-value < 0.05). Check practical significance (is the lift large enough to matter for the business?). Segment results: does the effect hold across platforms, geographies, user cohorts? Check for novelty effects (lift may fade as users get used to the change). Then: Ship, Iterate, or Kill.

Real Interview Cases

πŸ“‹ Solved Case Studies β€” Real Company Questions

These are real case study questions asked at product companies in India in 2024–2026, with complete structured answers you can adapt.

Swiggy
Funnel Drop-off Β· Product Analytics
Medium
Question
“Swiggy’s ‘Add to Cart’ to ‘Order Placed’ conversion rate dropped from 72% to 58% last week on Android. How do you diagnose and fix this?”

πŸ” Step 1 β€” Clarify

AIs the 14pp drop sudden (single day) or gradual (over the week)? β†’ Sudden = technical issue. Gradual = UX/product change.
BIs it only Android or also iOS and Web? β†’ Android-only = app release issue or OS-level bug.
CDid we release an Android app update this week? β†’ Yes β†’ High probability it’s the update.

πŸ“Š Step 2 β€” Segment

1By Android version: Check if drop is concentrated in specific OS versions β€” Android 14 vs older.
2By payment method: Is drop-off at payment selection or OTP step? Segment by UPI vs Card vs COD users.
3By city tier: Tier 1 vs Tier 2 β€” if Tier 2 only, could be network latency causing payment failures.

πŸ’‘ Hypotheses

H1New app update introduced a UI bug on the payment selection screen β€” UPI option not rendering correctly.
H2Payment gateway timeout increased β€” OTP delivery slowed β†’ users abandon before completing.
H3New “Add delivery instruction” step added β†’ extra friction in flow β†’ users drop at new step.

βœ… Recommendation

Immediate: Check Crashlytics/Sentry for errors in payment screen on Android. Check payment gateway latency dashboard. Roll back the feature causing new friction step if H3 is confirmed. A/B Test: Test the payment flow with vs without the new “delivery instruction” step β€” primary metric: checkout completion rate, guardrail: return rate. Success metric: Conversion back to β‰₯70% within 48 hours of fix deployment.

Flipkart
Revenue Analysis Β· Business Case
Hard
Question
“Flipkart’s GMV for electronics dropped 18% this quarter compared to last quarter despite overall traffic being flat. How do you diagnose and what do you recommend?”

πŸ”’ Revenue Decomposition

β†’GMV = Sessions Γ— Conversion Rate Γ— Average Order Value. Traffic is flat β†’ Sessions unchanged. So either Conversion or AOV dropped.
β†’Check: Conversion rate for electronics this Q vs last Q. Check: AOV for electronics this Q vs last Q.
β†’Suppose: Conversion flat, AOV dropped 18%. β†’ Users are buying lower-priced electronics. Why?

πŸ“Š Segmentation

1By sub-category: Smartphones (high AOV) vs Accessories (low AOV). Is smartphone sales mix dropping?
2External: Did Amazon/Croma run a competing sale this quarter that pulled high-value purchases away?
3Seasonal: Q4 often has sale events. Is this Q naturally lower (post-sale quarter effect)?

βœ… Recommendation

If AOV mix shifted: Launch targeted “upgrade nudge” for users browsing mid-tier phones β€” show premium features, EMI options prominently. If competitor price advantage: Real-time competitive pricing alerts for top 50 electronics SKUs. Metric to track: Smartphone + TV purchase rate (high-AOV categories) as leading indicator of GMV recovery. A/B test: EMI-first display on product pages β€” primary metric: AOV, guardrail: return rate.

Framework 6

πŸ“Š Metric Design β€” KPIs, Leading & Lagging Indicators

When asked to “design metrics for a new feature”, you need a structured framework β€” not a random list of numbers. Here is the complete approach.

1

Define the Feature Goal

What user problem does it solve? What business outcome does it drive? Example: “New onboarding flow for first-time users β€” goal is to increase Day-7 retention by getting users to their first value moment faster.”

2

Choose Your North Star (Primary KPI)

The one metric that tells you if the feature succeeded. For the onboarding example: Day-7 retention rate. This is the lagging indicator β€” it takes time to measure but is directly tied to the goal.

3

Define Leading Indicators

Metrics that predict the North Star before you can measure it. For onboarding: Onboarding completion rate, Time to first core action (e.g., first order placed), Number of features explored in session 1. These are measurable immediately and tell you if you’re on track.

4

Set Guardrail Metrics

Metrics you must NOT hurt. For onboarding: Session duration (don’t make it slower), Error rate (no bugs introduced), Customer support contacts (new flow shouldn’t be confusing). If the primary KPI improves but a guardrail degrades significantly β€” the change fails.

5

Define Success Thresholds

State what “success” looks like in numbers. “We’ll call this successful if Day-7 retention improves from 24% to 28% (16.7% relative lift) with no degradation in session duration (guardrail) β€” validated in an A/B test with 95% statistical significance.”

❓ Case Study Interview β€” Frequently Asked Questions
What is a North Star Metric?
+
A North Star Metric (NSM) is a single metric that best captures the core value a product delivers to users. It aligns all teams around one measure of long-term success. A good NSM reflects genuine user value (not vanity), predicts revenue, and is actionable by product, engineering and marketing teams. Examples: Spotify β†’ Monthly Active Listeners. Swiggy β†’ Weekly Orders per Active User. Airbnb β†’ Nights Booked. Never confuse revenue with NSM β€” revenue is a lagging outcome of delivering user value, not the value itself.
How do you approach a funnel drop-off case study?
+
Approach funnel drop-off in 5 steps: (1) Clarify the exact step, time period and platform, (2) Rule out technical issues β€” check if a new release coincided with the drop, (3) Segment the drop-off by platform, geography, user cohort and traffic source, (4) Form 2–3 hypotheses for the segment with the largest concentrated drop, validate each with session recordings, error logs and support tickets, (5) Recommend a specific fix with an A/B test β€” state primary metric, guardrail metrics, sample size and expected timeline.
What is the difference between conversion rate and retention rate?
+
Conversion rate measures the percentage of users who complete a specific action in a given session β€” e.g., 8% of visitors place an order. It is a single-session, single-event metric. Retention rate measures the percentage of users who return to use the product over a time period β€” e.g., 35% of users who installed last month are still active this month. Conversion is about getting users to take an action NOW. Retention is about keeping users coming back OVER TIME. Both are important β€” conversion drives acquisition revenue, retention drives lifetime value.
What are guardrail metrics in an A/B test?
+
Guardrail metrics are metrics that must not be negatively impacted when you make a product change, even if the primary metric improves. They protect the business from unintended consequences. Example: You’re testing a new checkout flow to improve conversion rate (primary metric). Your guardrail metrics might be: return rate (don’t want more impulse purchases that get returned), customer support contacts (new flow shouldn’t confuse users), and session duration (don’t want to make checkout artificially fast by hiding important information). If the primary metric improves but a guardrail degrades significantly, you do NOT ship the change.
How do you structure a data-backed recommendation in a case study?
+
Use the CRISP framework: (1) Context β€” what is the situation and the specific problem, (2) Root Cause β€” what the data shows is driving the problem (supported by segmentation), (3) Insight β€” the specific, non-obvious finding that changes how you think about the problem, (4) Solution β€” the specific recommendation with a clearly defined test to validate it, (5) Prediction β€” the expected impact in numbers and the timeline to measure it. Always end with a measurable commitment: “I’d expect conversion to improve by X% within Y weeks, which we’ll validate with an A/B test.” This shows you think in experiments, not opinions.
What is the difference between DAU, WAU and MAU?
+
DAU (Daily Active Users): Unique users who perform a core action on a given day. WAU (Weekly Active Users): Unique users active in the past 7 days. MAU (Monthly Active Users): Unique users active in the past 30 days. The DAU/MAU ratio (stickiness ratio) is particularly important β€” it measures how many of your monthly users come back daily. A ratio above 50% indicates a highly habit-forming product (WhatsApp ~70%). Below 20% suggests users find less daily utility. The appropriate window depends on your product’s natural usage frequency: daily apps use DAU, weekly use WAU, monthly subscriptions use MAU.

⭐ Key Takeaways β€” Business Case Studies

  • Always clarify before analysing β€” scope, time period, platform, baseline are essential to solving any case correctly
  • Funnel drop-off: Rule out technical first β†’ segment by platform/cohort/time β†’ form hypotheses β†’ validate β†’ recommend with A/B test
  • North Star Metric must reflect user value, predict revenue, and be actionable by all teams β€” never pick revenue itself as NSM
  • Revenue = Volume Γ— Price Γ— Frequency Γ— Retention β€” decompose before diagnosing
  • Retention is the most important health metric β€” fix retention before spending more on acquisition
  • A/B test design: Hypothesis β†’ Primary metric β†’ Guardrail metrics β†’ Sample size β†’ Duration β†’ Analysis β†’ Ship/Kill decision
  • Metric design: NSM (lagging) + Leading indicators + Guardrail metrics + Success thresholds = complete metric framework
  • Every recommendation must end with: “I’d validate this with an A/B test targeting X metric, expecting Y% lift in Z weeks”

Practice Case Studies with a Real Mentor

Our mock case study sessions simulate actual interviews at Swiggy, Flipkart, Amazon and more. Real questions, live structured thinking feedback, written report.

Book Free Case Study Session
PS
Prakhar Shrivastava
Founder, InterviewReady India Β· 10+ Years Product Analytics
Prakhar has conducted 500+ mock case study interviews and mentored analysts at Google, Flipkart, Swiggy, and top startups. He specialises in teaching structured analytical thinking β€” the skill that distinguishes great analysts from good ones.
βœ“ Expert Author Β· E-E-A-T Verified Β· Industry Experience