Amazon’s Senior Business Analyst (SBA) role is one of the most data-intensive BA positions in the industry. Unlike a standard BA role, SBAs at Amazon are expected to own end-to-end analytics — from writing complex SQL to building self-serve dashboards, defining business KPIs, and driving leadership-level recommendations using Amazon’s Leadership Principles as the north star. The interview process is heavily LP-driven with hard technical rounds in SQL and business metrics design.
This guide covers real interview questions from Amazon’s Retail, AWS, Advertising, and Logistics teams — with full answers, SQL queries, and LP-aligned response frameworks.
📚 Recommended Resources to Crack This Interview
2200 Most Asked Analytics Interview Questions
- 2,200+ questions · 23 topics
- For: those who want to master all analytics interview topics
Ace Any SQL Interview
- 220+ questions · SQL advanced, data models, DBMS
- For: anyone who wants to crack any SQL round (fresher to advanced)
Round 1 — SQL & Data Analysis
Amazon’s SQL round for SBAs is done in a shared SQL editor (Redshift or standard SQL). Expect 3 questions in 45–60 minutes. You’re evaluated on correctness, edge case handling, and whether you can explain your approach while coding.
Question 1: Prime Member Cohort Spend Analysis
You have orders (order_id, customer_id, order_date, total_amount, is_prime) and customers (customer_id, signup_date, country). Find the average order value (AOV) for Prime vs non-Prime customers, broken down by months-since-signup cohort (0–3 months, 4–6 months, 7–12 months, 12+ months). Return only rows where each cohort-prime segment has at least 100 orders.
WITH enriched AS (
SELECT
o.order_id,
o.customer_id,
o.total_amount,
o.is_prime,
DATEDIFF(MONTH, c.signup_date, o.order_date) AS months_since_signup
FROM orders o
JOIN customers c ON o.customer_id = c.customer_id
),
bucketed AS (
SELECT
CASE
WHEN months_since_signup BETWEEN 0 AND 3 THEN '0-3 months'
WHEN months_since_signup BETWEEN 4 AND 6 THEN '4-6 months'
WHEN months_since_signup BETWEEN 7 AND 12 THEN '7-12 months'
ELSE '12+ months'
END AS cohort_bucket,
is_prime,
total_amount
FROM enriched
)
SELECT
cohort_bucket,
is_prime,
COUNT(*) AS order_count,
ROUND(AVG(total_amount), 2) AS avg_order_value
FROM bucketed
GROUP BY cohort_bucket, is_prime
HAVING COUNT(*) >= 100
ORDER BY cohort_bucket, is_prime DESC;
- → How does Prime membership upgrade timing affect this analysis (customer becomes Prime mid-cohort)?
- → Add a column for median order value alongside average.
- → Which cohort would you prioritize for a Prime upgrade campaign, and why?
Question 2: Seller Performance Scorecard
You have seller_orders (order_id, seller_id, order_date, promised_delivery_date, actual_delivery_date, is_returned, refund_amount). Build a quarterly seller scorecard showing: on-time delivery rate, return rate, total refund amount, and a composite score = (on_time_rate × 0.5) + ((1 – return_rate) × 0.3) + (min(1, 1000/total_refund) × 0.2). Flag sellers with composite score below 0.6.
WITH quarterly_metrics AS (
SELECT
seller_id,
DATE_TRUNC('quarter', order_date) AS quarter,
COUNT(*) AS total_orders,
SUM(CASE WHEN actual_delivery_date <= promised_delivery_date
THEN 1 ELSE 0 END) * 1.0 / COUNT(*) AS on_time_rate,
SUM(CASE WHEN is_returned THEN 1 ELSE 0 END) * 1.0 / COUNT(*) AS return_rate,
SUM(COALESCE(refund_amount, 0)) AS total_refund
FROM seller_orders
GROUP BY seller_id, DATE_TRUNC('quarter', order_date)
),
scored AS (
SELECT
seller_id,
quarter,
total_orders,
ROUND(on_time_rate, 4) AS on_time_rate,
ROUND(return_rate, 4) AS return_rate,
total_refund,
ROUND(
(on_time_rate * 0.5)
+ ((1 - return_rate) * 0.3)
+ (LEAST(1.0, 1000.0 / NULLIF(total_refund, 0)) * 0.2),
4
) AS composite_score
FROM quarterly_metrics
)
SELECT
*,
CASE WHEN composite_score < 0.6 THEN 'AT RISK' ELSE 'HEALTHY' END AS status
FROM scored
ORDER BY quarter DESC, composite_score ASC;
NULLIF(total_refund, 0) prevents division by zero when a seller has zero refunds (which would otherwise give infinite score on the refund component). LEAST(1.0, ...) caps the refund component at 1.0 so sellers with very low refunds don’t get scores over 1.- → The composite score formula was set by leadership. What would you suggest to improve it?
- → How do you handle sellers who only had 5 orders in a quarter — should their score be trusted?
Question 3: Category Revenue Waterfall
You have sales (date, category, subcategory, revenue). Write a query to show, for each category, the revenue in 2024 vs 2025, the YoY change, and the percentage contribution of each category to total YoY growth. Sort by contribution descending.
WITH yearly AS (
SELECT
category,
SUM(CASE WHEN EXTRACT(YEAR FROM date) = 2024 THEN revenue ELSE 0 END) AS rev_2024,
SUM(CASE WHEN EXTRACT(YEAR FROM date) = 2025 THEN revenue ELSE 0 END) AS rev_2025
FROM sales
GROUP BY category
),
with_delta AS (
SELECT
category,
rev_2024,
rev_2025,
rev_2025 - rev_2024 AS yoy_change
FROM yearly
),
total_growth AS (
SELECT SUM(yoy_change) AS total_yoy FROM with_delta
)
SELECT
w.category,
ROUND(w.rev_2024, 0) AS revenue_2024,
ROUND(w.rev_2025, 0) AS revenue_2025,
ROUND(w.yoy_change, 0) AS yoy_change,
ROUND(w.yoy_change * 100.0 / t.total_yoy, 1) AS pct_contribution_to_growth
FROM with_delta w
CROSS JOIN total_growth t
ORDER BY pct_contribution_to_growth DESC;
Round 2 — Business Case & Metrics Design
This round tests your ability to define the right metrics, diagnose business problems, and design measurement approaches for Amazon’s scale. Leadership Principles (especially Customer Obsession, Dive Deep, and Insist on High Standards) should be woven naturally into your answers.
Question 1: Drop in Amazon Seller Satisfaction
The seller satisfaction score (CSAT) for Amazon Seller Central dropped 8 points QoQ. Your manager asks you to investigate. Walk through your full diagnostic approach.
Step 1 — Validate the signal. Before panicking: Did the CSAT methodology change? Did the survey timing shift (end-of-quarter surveys skew negative)? Is the 8-point drop uniform or driven by a single seller segment?
Step 2 — Segment the drop (MECE decomposition).
- By seller size: Large sellers vs SMBs — did the drop concentrate in one tier?
- By seller tenure: New sellers (enrolled < 6 months) vs established — onboarding issues vs operational?
- By category: Electronics, apparel, grocery — did a new policy change (returns, fees) hit a specific category?
- By geography: India, US, UK — was there a regional event (logistics delay, fee restructure)?
Step 3 — Identify likely drivers (hypotheses).
- Fee increase announcement — sellers often express dissatisfaction in surveys immediately after fee changes
- Increase in account suspensions or listing removals — frustration with policy enforcement
- Degraded Seller Central UI or slower support response time
- Competitor (Flipkart, Meesho) actively poaching sellers with better terms
Step 4 — Data sources to pull. CSAT verbatim comments (NLP cluster for top topics), support ticket volume and resolution time, account action logs (suspensions, listing removals), seller forum posts, QoQ seller churn rate.
Step 5 — Recommendation structure. Immediate: close-loop with the bottom 10% of sellers (proactive outreach). Medium-term: fix the top 2 operational drivers. Report weekly to leadership with a clear owner for each action item.
Question 2: Launch Metric Design for a New Feature
Amazon is launching a new “AI-powered restock recommendation” feature for sellers. Define the success metrics for this feature’s launch. Include primary metrics, guardrail metrics, and how you’d know if it’s working after 30 and 90 days.
Primary Success Metrics:
- Feature adoption rate: % of eligible sellers who viewed and acted on at least 1 recommendation within 30 days
- Stockout reduction rate: % change in stockout days per seller (treatment vs control cohort) — this is the core value proposition
- GMV lift: Revenue change for sellers who followed recommendations vs those who didn’t — did acting on the AI reco increase their sales?
Guardrail Metrics (must not worsen):
- Overstock rate — if recommendations cause sellers to over-order, that’s a failure even if stockouts drop
- Seller CSAT — a confusing feature that frustrates sellers is worse than no feature
- Recommendation accuracy (precision) — what % of “restock now” alerts were actually needed within 7 days?
30-day check-in: Adoption rate ≥ 15%, no degradation in CSAT, recommendation precision ≥ 60%, early stockout reduction signal in pilot cohort.
90-day full evaluation: Statistically significant GMV lift (≥ 2%), stockout days reduced by ≥ 10%, overstock rate unchanged, seller retention for feature users vs non-users.
- → Adoption is at 3% after 30 days. What do you do?
- → How do you design the experiment — what’s your randomization unit?
- → GMV went up 5% but overstock also went up 8%. Is this a success?
Round 3 — Leadership Principles (Behavioral)
Amazon’s LP round at the Senior level expects you to show strategic impact, cross-functional influence, and high-judgement decisions — not just “I did X and it worked.” Use the STAR format but go beyond execution: show how you challenged assumptions, influenced without authority, and dealt with ambiguity.
LP 1: Customer Obsession
Tell me about a time you went significantly beyond your job description to serve the customer (internal or external) — not because you were asked to, but because you identified a gap.
What the interviewer wants to hear:
- You proactively identified a customer pain point from data, not from a complaint
- You took ownership without being asked — this wasn’t your task
- The intervention was data-driven and measurable
- There was some friction (resource constraints, competing priorities) you navigated
Structure your answer to show:
- Situation: Describe the customer group and their pain point. Make it specific — not “customers were unhappy” but “sellers in the 50–500 GMV/month tier had a 34% higher support contact rate.”
- Task: Clarify that this was beyond your scope — you took this on independently.
- Action: Walk through the analysis → root cause → solution you built. Quantify effort.
- Result: Lead with the customer outcome first (support tickets reduced by X%), then business impact (cost saved, retention improved).
LP 2: Insist on High Standards
Describe a time when you raised the bar on the quality of analysis or reporting in your team — even when others thought the existing approach was “good enough.”
What the interviewer wants to hear:
- You identified a quality gap that others had accepted or normalized
- You pushed for the higher standard even when it created short-term friction
- Your push for quality improved something measurable (fewer errors, better decisions, more adoption)
- You maintained relationships while holding your ground
Anti-patterns to avoid:
- Don’t make it sound like you were just being difficult or pedantic — there must be a business reason the higher standard mattered
- Don’t say “the team finally agreed with me” without explaining how you brought them along
- Don’t position yourself as the sole hero — mention how you enabled others to meet the higher bar
- → What was the pushback you received, and how did you handle it?
- → How did you ensure the higher standard became the new default, not just a one-time thing?
- → What would you have done if leadership had decided the current standard was acceptable?
4-Week Prep Plan for Amazon Senior Business Analyst
Build SQL foundation (window functions, CTEs, CASE aggregations), refresh Amazon’s 16 Leadership Principles with real examples mapped to your experience, and read TDM’s interview pattern posts on Amazon BA and analytics.
Work through the 2200 Most Asked Analytics Interview Questions ebook — cover SQL, business case, and KPI design sections. At the end of Week 2, book a mock interview at topmate.io/nitin_kamal to stress-test your LP answers.
Complete the Ace Any SQL Interview ebook (link) — all 220 questions. Focus on Redshift/PostgreSQL syntax: DATEDIFF, DATE_TRUNC, NULLIF, window functions with RANGE vs ROWS. Book second mock interview.
Prepare 2–3 STAR stories for each of the 6 most common Amazon LPs (Customer Obsession, Dive Deep, Ownership, Insist on High Standards, Bias for Action, Deliver Results). Timed practice: 2 LP stories + 1 SQL problem per day. Final mock interview at the end of Week 4.
Target: ~950 questions + 3 mocks + full LP bank = ready for Amazon’s bar raiser round.