Product design question
Design a product to recommend items to customers for purchasing via Amazon Prime Now, with the objective of optimizing the number of stock keeping units (SKUs) and warehouse space.
- Amazon
- 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 product design under an explicit operational constraint (SKU and warehouse space), not just a feature wish list, which is unusual for a recommendation design question.
How to approach it
- Clarify the goal: Prime Now is fast (often same/next-day) delivery from local micro-fulfillment centers, so recommendations must respect what is physically stocked nearby, not Amazon's full catalog.
- Identify the constraint's implication: recommendations should favor high-velocity, space-efficient items over long-tail SKUs, since warehouse space is scarce and each SKU has a holding cost.
- Design the recommendation logic: surface items with high purchase probability per user (based on past orders and local demand) but weighted toward SKUs with strong sell-through, so slow movers are recommended less and can be phased out.
- Add a complementary-item layer: recommend items frequently bought together (like a produce basket suggesting a recipe's missing ingredients) to raise basket size without adding new SKUs.
- Feed this signal back into inventory planning: use recommendation click and purchase data as an input for which SKUs to keep versus cut from warehouse space.
- Define success as basket size per order and, separately, SKU turnover rate (inventory turns), which is the operational metric this design is meant to improve.
What a strong answer includes
- Explicitly closes the loop between recommendations and inventory: recommendation data should inform which SKUs get warehouse space, not just drive sales.
- Distinguishes personalized recommendations (based on a user's history) from local-availability constraints (what the nearest facility actually stocks).
- Gives a concrete mechanism: complementary-item bundling (recipe-based suggestions) as a way to grow basket size without adding new SKUs.
- Names two separate metrics, basket size and SKU turnover, since the question asks for both revenue and space optimization.
Common mistakes
- Designing a generic 'recommend similar items' feature that ignores the SKU and warehouse-space constraint entirely.
- Optimizing purely for basket size while ignoring that expanding SKU variety fights the space-optimization goal.
- No feedback loop from recommendation performance back into inventory decisions.
Likely follow-up questions
- How would you decide which slow-moving SKUs to cut first?
- How would this recommendation logic change between a dense urban facility and a suburban one?
- What would you do if a recommended item was frequently out of stock locally?
More product design questions
- Redesign Twitter for eCommerce.Shopify · Product design · Hard
- Design a loyalty program for a search engine that's competing with Google & Bing.Google · Product design · Hard
- For a retail store, users are frequently complaining about long queues during check out. What strategy would you use to improve the experience in the short term and long term?Shopify · Product design · Medium
- Give an example of a badly designed product.Adobe · Product design · Easy
- Design a library for the future.Spotify · Product design · Hard
- Explain some new Alexa Skillsets you would roll out and why.Amazon · Product design · Medium
More questions from Amazon
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