Lars Damgaard
strategic user experience designer
August 8th 2013

The broken information architecture of Netflix content exploration. Or why multiple categories suck

When browsing for content in Netflix, I have always noticed that the same titles appear again and again across categories.


This is annoying for several reasons. First of all because browsing categories is the Netflix interface per se. Secondly because categories tend to loose their meaning if they are not distinct. As a result, it becomes futile to browse through the massive amount of categories.


Finding content on Netflix: a personal use case

Regardless of the device being used, the main purpose of Netflix is to find good titles of interest and watch them. As simple as that.

Based on my own use, I guess five strategies for browsing content on Netflix can be identified:

I find social, suggestion and relatedness to be funny and interesting ways of exploring content, but more often than not, I want to find titles of a certain kind (browsing by genre). Or perhaps even more often, I want to see if anything new has arrived since the last time I checked (browsing by recency). This is where Netflix fails in my opinion.

Categories are tags rather than categories

For example, when browsing by recency, you have two options: you can either look through a category labelled new or a category called just added. This is weird for two reasons. First of all, you have to interpret the difference between new and just added. That’s pretty difficult! Secondly, like I wrote in the introduction to this post, you will soon realize that titles appear again and again across categories: the category labelled independent films contains many films that also appear in the category labelled drama and so on. As a result, it feels almost futile to browse through categories, simply because they are not categories.

This is the fundamental challenge of the Netflix taxonomy model: in the underlying datamodel, categories are probably tags and not categories if we define categories as distinct content structures.

Tags are good because they fit reality

At a glance, using multiple tags instead of distinct categories is brilliant. It serves the complex reality of films, where very few films belong in one category only. Furthermore, it probably makes it a lot easier to develop algorithms for suggestions and relatedness (which is the stuff that Netflix does really, really well).

Category browsing as complexity reduction?

However, it does not work very well when it comes to the main part of the Netflix content exploration interface: category browsing. Most of the interaction elements we provide users with when we want them to explore content is there to reduce the complexity of a large, unfiltered resultset. Filters in a faceted navigation are the most famous pattern, but category browsing isn’t any different.

In the case of Netflix, the user cannot be sure that her effort to look through categories will be worthwhile. The reduction of complexity is missing. Or even worse, the main interface of category browsing in Netflix systematically produces complexity and confusion in the relationship between titles and categories.

On purpose?

Perhaps it’s an unintended consequence of a tag based taxonomy model; perhaps it’s a deliberate attempt of manipulation in which Netflix wants it’s users to feel that the amount of content is bigger than it actually is.

Alignment of information model and the user interface design

Regardless of the reason, I think the interface for content exploration should be aligned with the underlying datamodel and that does not seem to be the case for Netflix.

Thanks for reading.

Related to what you were just reading
June 9th 2014

Can the digital IKEA effect increase user engagement and brand loyalty?

According to behavioral scientists Michael I. Norton, Daniel Mochon and Dan Ariely, people who participate in the creation of something tend to evaluate what they have created more positively than they would have done if they had not…

Read more