Measure TV Show Addictiveness - Product Metric Design Interview

Question

## Problem

You are a data scientist at a streaming platform. Define and compute a metric for "addictiveness" of a TV show -- how likely users are to keep watching after starting.

You have:
```
view_events(user_id, show_id, episode_num, watch_start, watch_end, session_id)
shows(show_id, title, total_episodes, release_date)
```

**Part 1:** Propose a formula for an addictiveness score. Justify each component.

**Part 2:** Write a SQL query to compute your score for all shows with at least 100 unique viewers.

**Part 3:** A show's score drops sharply at episode 3. List 3 hypotheses and the SQL or analysis you would run to investigate each.

## Example metric

One reasonable definition: `completion_rate = avg(max_episode_watched / total_episodes)` per user who watched at least episode 1. Weight by binge depth: avg consecutive-episode sessions per user.

## Follow-ups
- How do you handle shows still airing (total_episodes unknown)?
- What confounders could inflate addictiveness for shorter shows?
- How would you A/B test whether showing episode previews improves the score?

Measure TV Show Addictiveness - Product Metric Design Interview

Question Details

Full Details

Topics

More from Bloomberg