Netflix — Software Engineer ❌ Failed

Level: Unknown Level

Round: Phone Screen · Type: Coding · Difficulty: 6/10 · Duration: 60 min · Interviewer: Unfriendly

Topics: Data Analysis, User Engagement

Location: Los Gatos, CA

Interview date: 2025-08-15

Got offer: False

Summary

I had a technical phone screen. I was asked to analyze user engagement patterns across different content. The goal was to classify viewing sessions and determine each user's most common engagement pattern.

Details

The coding question I got was:

`python

We're analyzing user engagement patterns across different content.

A user's engagement with a piece of content is defined by:

- "Abandoned" if they watched less than 25% of the content

- "Sampled" if they watched 25% to 75% of the content

- "Completed" if they watched more than 75% of the content

Given a list of viewing sessions where each session contains:

- account_id

- title_id

- title_runtime_secs (total length of the content in seconds)

- watched_secs (how much the user watched in seconds)

- day_nbr (an integer representing the day of viewing, where consecutive numbers are consecutive days)

Write a function that:

1. Classifies each session as "Abandoned", "Sampled", or "Completed"

2. For each user, finds their most common engagement pattern across all their sessions

3. Returns a dictionary with the count of users for each dominant engagement pattern

viewing_data = [ {"account_id": "U1", "title_id": "T1", "title_runtime_secs": 3600, "watched_secs": 3500, "day_nbr": 1}, {"account_id": "U1", "title_id": "T2", "title_runtime_secs": 5400, "watched_secs": 5000, "day_nbr": 3}, {"account_id": "U1", "title_id": "T3", "title_runtime_secs": 7200, "watched_secs": 6800, "day_nbr": 6}, {"account_id": "U2", "title_id": "T1", "title_runtime_secs": 3600, "watched_secs": 3500, "day_nbr": 5}, {"account_id": "U2", "title_id": "T4", "title_runtime_secs": 2700, "watched_secs": 2600, "day_nbr": 9}, {"account_id": "U2", "title_id": "T5", "title_runtime_secs": 4500, "watched_secs": 4300, "day_nbr": 10}, {"account_id": "U3", "title_id": "T2", "title_runtime_secs": 5400, "watched_secs": 5200, "day_nbr": 2}, {"account_id": "U3", "title_id": "T3", "title_runtime_secs": 7200, "watched_secs": 6500, "day_nbr": 3}, {"account_id": "U3", "title_id": "T4", "title_runtime_secs": 2700, "watched_secs": 2000, "day_nbr": 4}, {"account_id": "U4", "title_id": "T1", "title_runtime_secs": 3600, "watched_secs": 2000, "day_nbr": 1}, ]

def analyze_engagement_patterns(viewing_data): TODO `

My approach:

1. First, I parsed the viewing data to classify each session into "Abandoned", "Sampled", or "Completed" based on the watched percentage.
2. Then, I aggregated the engagement patterns for each user.
3. Finally, I determined the most common engagement pattern for each user and counted the number of users for each dominant pattern.

Key insights:

• Handle edge cases such as zero runtime to avoid division by zero.
• Use appropriate data structures (e.g., dictionaries) to efficiently aggregate and count engagement patterns.