With 209 million subscribers worldwide (as of Q2 2021), Netflix is the undisputed best video streaming platform. Its main source of revenue is subscribers’ subscription fees, so it has become Netflix’s most important goal to continuously push titles that users are interested in to improve user retention. To do this, Netflix must give users a highly personalized viewing experience. In this regard, Netflix once said: “Netflix is not a product, but hundreds of millions of products tailored for every user. 」(“Personalization … enables us to not have just one Netflix product but hundreds of millions of products: one for each member profile.”) Here’s an analysis of how Netflix can make you want to watch it through a personalized experience!

Why is personalized experiences important?

From video platforms to social media to online stores, how to capture users’ attention is a major issue. The more time a user spends browsing content, the more ads they receive and the more products they buy. The key is to personalize the experience and capture the user’s attention by recommending content that interests them, so they can spend more time on the platform. People have very short attention spans — studies have shown that if Netflix doesn’t recommend a title within 60 to 90 seconds, users will lose interest and leave Netflix!

How personalized is the Netflix recommendation system?

From the content and arrangement of the videos on the homepage, to the details of the videos, to the emails and user notifications, Netflix uses different algorithms to provide users with information about the works they are most interested in and the videos they care about the most. Netflix’s homepage categorizes recommended titles by theme and genre in up to 40 rows. As shown in the figure below, each row is a film theme or genre, such as “Blockbuster Movie”, “Parent-Child Movie”, “Psychological TV Drama”, etc.

Image credit: Netflix

● What video themes should be included on the homepage? (For example, would the user be interested in a “psychological TV drama” or a “friendship-related award-winning show”?)

● Which topics should be at the top of the list?

● What videos should be included in each topic? How should the videos be sorted? (For example, is “Aquaman” or “Wild Speed” at the top of the “blockbuster movie” theme?)

All of the above is carefully calculated by the video recommendation system, and the purpose is to find the video they want to watch in the shortest time after logging in. And Netflix’s arrangement of movies in rows by category also has an ingenious idea. In addition to systematically categorizing videos to facilitate viewers’ selection of videos, this is also for more effective analysis of user behavior. When the user scrolls down, he is not interested in the topic on the screen, and when he scrolls left, he is interested in the topic being presented, but not in the works that rank in the topic.

A set of films, 9 posters

In addition to the homepage, the detailed information of each video is tailored to the user:

● Which version of the video introduction and highlights should be shown?

● How should the video be labeled?

● Which version of the poster should be displayed? (Netflix calls it “artwork”)

Netflix has edited a total of 10 versions of House of Cards based on the characteristics of different audiences, and Stranger Things alone has 9 posters that users are likely to see! Each poster of the film shows what makes the work unique, whether it is a close-up of the famous actors who have participated in the work, a classic scene from the work, or a scene that best expresses its essence. Which poster is shown depends on the user’s past behavior data and interests. For example, if you’ve watched action/adventure movies in the past, Netflix will choose a poster version that includes action or thrill scenes when it comes to new titles.

Image source: netflixtechblog.com

Collect user behavior data in real time

To personalize the experience, Netflix needs to collect a lot of user data at any given moment. The video recommendation system continuously changes and updates the works and related materials it pushes to users based on various user usage data. The data collected by Netflix includes:

● Video content and search content that the user has clicked on in the past

● The user’s interaction with the video – including pacusing, rewatching or leaving the video, and watch time

● User’s viewing habits – including days (weekdays or weekend?) , time slot and device (phone, tablet or laptop?) wait

● Video content that the user didn’t click on (i.e. wasn’t interested).

● Behavioral data of other users who share the same interests and behaviors as you

Every seemingly small action of a user on the platform is extremely significant to Netflix! In addition, the newer the behavior data, the higher the proportion of the system calculation process.

If you click on the video, you’ll see a score below the play button that says “X% is right for you”, which is calculated by the Netflix recommendation system to match the video’s interests:

Image credit: Netflix

Even for the same video, each user’s “fit” score is unique. Because each user has different interests and viewing habits, a movie may be “91% suitable” for User A, but “71% Suitable” for User B.

Encourage user feedback to improve system accuracy

In addition to continuously collecting user behavior data, Netflix also actively encourages user feedback so that the recommendation system can better understand each user’s interests and preferences. To the right of the playback button, there’s a Thumbs Up (for “I like this video”) and a Thumbs Down (for “Not to My Taste”) buttons to let Netflix know if you’re not interested in the video and update its recommendation system.

Image credit: Netflix

Netflix’s ability to deliver the ultimate personalized viewing experience in every part and detail of the platform is based on its constantly updated user database and algorithmic algorithms. Today, 80% of Netflix users’ hours of viewing come from its movie recommendation system, which can be said to truly achieve “the right content to the right person”. Data analytics plays a crucial role in this, and Netflix has used big data to successfully buy hits around the world (such as “House of Cards”), and we will be writing a separate article to share with you in the future.

YouFind offers a full range of digital marketing solutions, including search engines, online advertising, social media, and more. We will update you with the latest marketing information on a regular basis.