Product design question
Develop a seller score algorithm for Walmart Marketplace.
- Walmart
- Product design
- Hard
Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.
Start a mock interview on this question · Mock interview from a job description
What this question tests
Tests designing a scoring system from a PM lens: picking input signals, weighting logic, and gaming guardrails, not writing the model itself.
How to approach it
- State the goal: help buyers trust sellers and help Walmart promote, flag, or remove sellers appropriately.
- List inputs: on-time shipping rate, defect/return rate, review rating, message response time, and policy violation history.
- Weight by risk: defects and violations weighted heaviest since they harm buyer trust and Walmart's liability; responsiveness weighted lightly.
- Weight recency over history: score last 90 days more heavily, and require a minimum order volume before a score is trusted.
- Guard against gaming: sellers could inflate reviews or cherry-pick orders, so add anomaly detection and audits for suspiciously perfect scores.
- Define use and success: tiered outcomes (search boost, warnings, suspension), validated against actual buyer satisfaction over time.
What a strong answer includes
- Weights inputs by real business risk (defects, violations) rather than treating all signals equally.
- Addresses cold-start fairness explicitly with a minimum order volume before a stable score, a commonly missed detail.
- Names concrete gaming risks and proposes anomaly detection, recognizing any visible score will be gamed.
- Defines how the score drives real consequences and validates it against actual buyer outcomes.
Common mistakes
- Proposing a score with no discussion of what happens as a result of it.
- Ignoring gaming risk, a first-order concern in any reputation system.
- Weighting all inputs equally with no reasoning about which matter most.
Likely follow-up questions
- How would you handle a seller disputing their score as unfair?
- How would you stop large sellers from skewing average scores?
- How would you communicate score changes to drive improvement, not just punish?
More product design questions
- Redesign Twitter for eCommerce.Shopify · Product design · Hard
- How can Google help improve the grocery shopping experience of users?PayPal · Product design · Easy
- Design a Supermarket for older people.Google · Product design · Medium
- How would you improve Facebook for commuters?Walmart · Product design · Medium
- How would you improve Uber?PayPal · Product design · Easy
- How would you improve Amazon?Amazon · Product design · Easy
More questions from Walmart
Learn the skill behind it
Chapters of the AI PM course that teach what this question tests.
- Chapter 4: Discovery and strategy for AI products
- Chapter 7: AI UX and human oversight: design for a system that is wrong sometimes
- Chapter 14: Get the job: the AI PM interview loop