How Netflix Prioritizes and Classifies Content Projects

Purpose of This Research

To identify the processes and internal systems that drive the priorities and classifications of projects at major streaming services:

• ‍Assessing new title production and renewal processes
• Identifying key data points for decision making
• Highlighting factors internal teams rely on

‍

‍

Netflix is a digital streaming service that has three main areas of concern in project orientation.

• ‍Show & Movie Purchasing. Licensing existing content is an upfront investment with lower stakes than content creation but equal importance for data collection and understanding user interests.
• ‍Original Content Creation. When making decisions about the production process, existing data in combination with platform usage big data function as main collection and information points.
• ‍Project Renewal. Once the leap has been made to license or create an original project, Netflix must decide if they will renew or cancel those assets.

‍

Netflix’s release strategy supports an anti-windowing (1) approach. Content is disseminated in an accessible way, all at once.

This is considered a binge-forward release strategy where content for an entity becomes entirely available.

‍

When focusing on new content, Netflix team members are led by two main questions:

• Which existing titles are comparable and in what ways?
• What audience size can we expect and in which regions?

‍

Netflix uses machine learning, specifically transfer learning, which is built of a target task and a source task to answer their leading content decision questions.

• Target Task
• Source Task

‍

Source tasks compile a large sample of three main elements for their analysis and run them to generate…

Source (2)

‍

…a specific sample for target tasks that are relevant for content making decisions.

Source (2)

‍

Knowledge graphs are built with assessment factors that create connections to predict viewer interests and success.

• ‍Based On: What is the origin of an existing title?
• ‍Storyline: What are the main morals or plot points?
• ‍Primary Era: When does it take place?
• ‍Genre: What genre is it?

‍

These are formed with analysis from machine ‘transfer’ learning that build relationships to…

Source (2)

‍

…create a network that can be used to help inform decisions, such as this simple example (3) with “Orange is the New Black”.

Source (2)

‍

Machine learning and analytics play a major role in marketing decisions for Netflix. All company decisions are run on A/B testing (2).

From product development, to title covers and trailers, machine learning plays a role in categorizing and identifying top performing assets & strategies. 

Source (2)

‍

Content is being filtered through machine learning systems to identify context of scenes for personalization in marketing assets.

Source (2)

‍

Combining data compiled from machine learning with A/B testing, successful campaigns can be drawn on for decision making.

See how the same movie title cover crossing two genres can be adapted based on precedent.

Source (2)

‍

Netflix uses a viewer classification (4) system to note how users interact with individual pieces on content on the platform. 

Classification data processes viewership during first 7 days of a premier and after 28 days of launch.

‍

Digital I, a British SVOD data company, released Netflix’s original “First Kill” completion data following news of its cancellation in August 2022.

The show had under 45% (5) of people who started episode 1 and watched all the way through to episode 8.

‍

References:

1. Netflix Says It’s Committed to a Binge-Release Strategy for Series

‍2. Supporting content decision makers with machine learning

‍3. How Netflix Uses Analytics To Select Movies, Create Content, and Make Multimillion Dollar Decisions

‍4. This is how Netflix classifies us as a user: 'starter', 'observer' or 'completer'

‍5. How Does Netflix Decide Whether to Renew or Cancel a Show?