TikTok · AI product · Senior · 45 to 60 minutes

How should TikTok design video discovery?

TikTok is a mobile app where people watch and create short videos recommended to them. Answer this product management case and get it graded line by line against a weighted rubric.

Product strategy
AI product judgment
Experiment design
Systems thinking
Executive communication

The situation

TikTok is preparing its next phase of discovery product development. The Head of Product for discovery asks you, a Group PM, to recommend how the primary video browsing experience should work for new and returning users. The team agrees that recommendation quality will matter, but there is disagreement about the interface that should anchor it. One camp wants a highly immersive viewing flow with very little else on screen and a simple way to move to the next clip. Another wants a more familiar browsing experience with more visible navigation, context, and nearby choices so users can compare, backtrack, and steer more directly. The decision cannot wait because creator supply is rising, short video competition is increasing, and the company is about to commit major engineering and design resources for the next two quarters. This is hard because the choice will affect first session engagement, user learning curve, content diversity, creator exposure, and what the recommendation system can improve over time. You need to recommend a direction, how to test it, and what risks TikTok should manage if it commits.

What to produce

A 2 page product strategy memo.

Your answer must

Describe 2 to 3 credible product directions for the primary discovery experience and the decision criteria you will use.
Make a recommendation and justify the tradeoffs across user experience, personalization quality, creator ecosystem effects, and business impact.
Define an experiment and measurement plan with leading and lagging metrics for new user activation, session depth, satisfaction, and content diversity.
Identify at least 3 second order effects or risks, and propose mitigations or guardrails for each.
Provide a rollout sequence for the next 2 quarters, including key cross functional dependencies across product, design, engineering, and ML.

Constraints

Must preserve a fast, mobile first short video experience.
Cannot assume users will follow creators or build a social graph before getting value.
You can change the browsing interface and ranking logic, but not launch a separate standalone app or creator program.

Format

Use sections for decision framing, options, recommendation, experiment plan, risks, and rollout.

How TikTok actually played it

Explains how TikTok's interface is designed around training its recommendation algorithm, with full-screen single videos that give clear signals. Shows how UI and ML co-design create a virtuous circle.

Design the interface to produce clean training signals
Constrain choices so the algorithm learns faster

Built from Seeing Like an Algorithm by Eugene Wei.