Ishita Ghosh

1. Overview

This is a case study of Netflix, an on-demand Internet streaming media service. It is certainly interesting to see how Netflix organizes their media so as to make their content more “discoverable” for their users. Therefore, this case study will focus on Netflix’s organizing of their media from the customer’s perspective, such that they are better able to discover, sort, and interact with the media resources, which in turns renders this process as seamless at the user end. Netflix separated its digital content from its physical DVDs service a year ago, although it must be noted that titles that are available on DVD will still show up on the search bar in Netflix’s online portal.

2. What resources are being used?

The resources in question are the different types of media that broadly includes movies and television content. These resources may already be organized to some basic extent based on simple, explicit tags such as language, basic genre (such as comedy, drama, romance, but not more complicated categories that Netflix eventually generates such as cerebral dramas) etc. The current scope and scale of Netflix’s organizational system seems dependent on their individual agreements with various production houses, which in turn determines what resources become digital content that can be streamed on-demand, online, versus what resources are stored onto physical DVDs (these services are now separated).

3. Why are the resources organized?

The resources are being organized to ensure easier and smoother “discoverability” by Netflix users, that not only reflects the popular tastes of Netflix users at large, but more specifically and perhaps more persuasively, reflects the user’s own personal tastes that are determined through a combination of review and rating mechanisms. Eventually, Netflix’s goal is to provide a seamless media viewing experience for its users, without too much exertion on the part of their users. Therefore, organizing the media resources in an optimal manner (irrespective of Netflix’s actual database size, as long as it remains relevant to the users), that lets users find media that is suited to their preferences, will ensure that they keep coming back to the platform to continue viewing Neflix’s media. To this end, Netflix supports different types of interactions. These interactions include searching, viewing, and reviewing on the part of the users. More specifically, the user can find media explicitly using the search bar on the top right (this is when they already know the title(s) of the media they want to  view) or they can discover media using Netflix’s pre-sorted genres (such as classics, comedies, cult movies etc) or through a more personalized feature that takes the user’s preferences into consideration (the account user will be greeted with captions such as because you watched 30 Rock and then a host of content similar to this will be presented; or be shown certain categories exclusively, such as cerebral dramas, based on a user’s tastes, trajectories, as well as the completion of ongoing surveys).

4. How much are the resources organized?

Evidently, Netflix has the capacity as well as the infrastructure to implement really complex organizing. The same resource may be organized across different categories. Broadly speaking, Netflix sorts these resources into basic, familiar genres, and may further add a level of granularity to these (such as certain resources under dramas may be further categorized into cerebral dramas). The same resource might then find itself into a “You may also like” categorization based on the user’s personal preferences. Eventually, Netflix’s goal remains to provide a seamless searching and viewing experience for its users. Broadly speaking, Netflix achieves this by a general sorting based on the resources’ descriptive properties (what basic genre does it belong to? what language?), but adds further granularity to this based on esoteric tagging, user reviews and ratings, as well as their viewing histories. Certainly, the extent of resource organization for a given user will also depend on how much that user views, as well as rates and reviews, the digital content on Netflix.

5. When are the resources organized?

The resources are being organized on two levels for users to better discover and access them. At one level, Netflix is continually employing the services of its in-house team as well as a small, esoteric, external team of “taggers” (such as documentary filmmakers for instance) to continually organize Netflix’s media content into pre-determined or emerging genres. However, Netflix also organizes at another level for the user, which is based on their personal preferences. This organization happens as and when the user reviews, rates, and views existing media. This seems the most feasible solution in order to accommodate both highly active users (they will have more tailor-made recommendations) as well as more inactive users (their personalized preferences might be lacking, but at least Netflix can offer them some basic recommendations).

6. Who does the organizing?

As mentioned previously, Netflix has an in-house team of “organizers” as well as an external team of “taggers” (who are typically movie experts in their own right) who essentially tag Netflix’s digital content and sort into pre-determined as well as emerging categories. Furthermore, ongoing ratings and reviews by the general corpus of Netflix users determines what is popular, leading to yet another categorization that is displayed prominently on the top of the homepage, “Popular on Netflix”. Netflix also uses automated processes that are able to determine, sort, and articulate this crowdsourced information. Therefore, through a combination of people as well as computation processes make the decisions about the organizing system; this makes the scale of Netflix’s vast digital content as well as user-generated information manageable, especially as it is organized by elected people, automated algorithms, and by the bottom-up labors of its audience. This is certainly preferable to just restricting the organizational activities to people (too slow) or to computational processes (devoid of crowdsourced/user preferences).

7. Other considerations

Currently, Netflix hasn’t attempted to integrate social media into their existing model. This may offer a new peer-review organizational capability to their existing organizational structure that may be immensely valuable for users. If computational processes can determine peers with similar tastes, even inactive users can be privy to more salient and better tailored recommendations, undoubtedly an attractive proposition.