Amazon product sense interview questions

Sample questions with short answers

Tap a question to reveal the short answer and key points.

1. 1Write the press release for a new Amazon feature (working backwards). What goes in it?MediumAmazon-specificShow short answer

   Short answer

   Working Backwards starts from the customer: a headline, the problem in the customer's words, the solution and why it's better, a quote from a leader and a customer, and how to get started. Then an FAQ with the hard questions — cost, risks, metrics. Interviewers check that every sentence is about customer benefit, not technology.

   Interviewers listen for

   • customer problem in their words
   • headline and solution benefit
   • FAQ with risks and metrics
2. 2Prime delivery promise is being missed in one city. How do you prioritise fixes?HardAmazon-specificShow short answer

   Short answer

   Quantify: what share of orders miss, by zone, carrier, and time slot. Separate demand spikes from capacity and routing issues. Prioritise by customer impact and fix cost — e.g. adjust promise windows for the worst zones immediately while you add capacity. Define the metric (on-time %) and guardrail (cost per package).

   Interviewers listen for

   • quantify misses by zone, carrier and time slot
   • adjust promise windows now, add capacity later
   • on-time metric with cost guardrail
3. 3How would you improve a product you use every day?MediumAsked everywhereShow short answer

   Short answer

   Pick a product, state its goal and the user segment you will focus on, then list their pain points and prioritise one by reach and impact. Propose two or three solutions, pick one, and define a success metric plus a guardrail metric. Call out the trade-offs and how you would test it. The structure matters more than the specific idea.

   Interviewers listen for

   • define goal and user segment
   • prioritise pain points by impact
   • success metric and guardrail metric
4. 4A key metric dropped 15% week over week. How do you investigate?MediumAsked everywhereShow short answer

   Short answer

   First check data integrity — tracking changes, logging bugs, timezone or attribution shifts. Then segment: platform, geography, new vs returning users, release version. Correlate with internal events (launches, experiments, pricing) and external ones (holidays, competitors, outages). Form a hypothesis, validate it, and only then propose a fix.

   Interviewers listen for

   • rule out data or tracking issues first
   • segment the drop
   • correlate with internal and external events
5. 5How do you decide what to build next when engineering capacity is limited?HardAsked everywhereShow short answer

   Short answer

   Anchor on the product goal for the quarter, then score candidates on user value, business impact, confidence and effort (RICE or a simpler impact/effort grid). Protect a slice of capacity for tech debt and reliability. Explain the trade-offs to stakeholders with the data behind the ranking, and revisit when new evidence arrives.

   Interviewers listen for

   • tie to the quarter's goal
   • score by impact, confidence and effort
   • reserve capacity for debt and reliability