|
Contributors | Messages | Polls | Resources |
|
How Netflix Learned to Read Your MindNetflix Inc. (Nasdaq: NFLX) is working to develop such a detailed predictive understanding of each of its users, that in the future they won't have to browse listings at all; the right video would automatically start playing the moment they open the app. According to Netflix's vice president of innovation, Carlos Gomez-Uribe, this is a perfectly realistic vision. In an interview with Business Insider UK earlier this week, he described the effort Netflix has made in developing effective ways to understand what its users like to watch. Huge quantities of usage data are gathered from 83 million subscribers worldwide, who are streaming billions of hours of video. The company then uses machine learning to spot patterns in this data, identifying similarities between individual users and groups of users to make predictions. Complex algorithms then use the data to identify the best titles for particular households and the best way to display them on the user's home page. And this works today: Netflix reports that 80% of titles watched today are already based on its recommendations. Netflix relies on actual usage data, which gives the company far more reliable insight than anything surveys or customer feedback have been able to provide. Netflix's analysis also adjusts for varying selection behaviors on different days of the week (for example, more impatient selection on weekdays, with shorter viewing sessions; and more leisurely on weekends). Separate research from video technology vendor Ooyala recently confirmed these video consumption variances over the course of the week. (See OTT 'Power User' Habits Analyzed.) Netflix has found that presenting various selections of titles in rows is the most effective way to display videos. Each row is based on a different algorithm, such as videos users are likely to return to (based on how long they watched the abandoned video, what they've watched since and how long ago they abandoned it). Other rows are based on what they watched or liked in the past, what genres they have watched, shows that are trending and popular and match previous viewing patterns, etc. The relative importance of each row (i.e., its rank in the list of rows from top to bottom on the home page) is based on yet another algorithm, which optimizes the rows that have been most effective in attracting users. Netflix will also customize artwork and even trailers for new shows to better suit different user profiles. The company says it saves $1 billion annually as a result of these measures. Firstly, this analysis helps Netflix quickly root out videos no one watches so it's able to get maximum value from its licensed catalog. It also helps in selecting the right titles to license in the first place. So according to the company, its catalog doesn't need to be as large, because it can better identify the right titles for its user base. This is critically important, because users typically get frustrated if they can't find a video that interests them in 60-90 seconds, usually having glanced through just 10-20 titles. And if that happens too many times, they will stop subscribing. Netflix also uses these insights to commission its own shows. The company is expected to spend $1 billion in 2016 on original programming, and ramp that up to $5.5 billion in 2020. That's a sizeable bill, and would require sustained subscriber growth worldwide to fund it. Content is a risky business -- hit driven and high risk -- and this kind of investment requires a very high hit ratio in order to be sustainable. Netflix is betting big on being able to deliver a string of hits, and this kind of user insight is at the heart of the company's gamble. (See Netflix to Double Down on Original Programming, Says UBS.) There is reason for the company to be confident. Netflix can identify the specific episode and sometime the exact plot event that got its users hooked on to a show. By that, Netflix means that 70% of users then went on to watch the whole series. Key moments this year such as the suspenseful episode two of Stranger Things, the murder in episode two of The Fall, or the interrogation of Brendan Dassey in episode four of Making a Murderer, are when Netflix users committed to those shows. In fact, Netflix has released a list of shows citing the particular episodes where its viewers got hooked. Take a look at the list below, and try to recall when it was that Netflix first got inside your head.
— Aditya Kishore, Practice Leader, Video Transformation, Telco Transformation |
Contentious issues that are likely to fuel lawsuits and angry blogs in the coming year.
Content producers are unhappy with the advertising approach and revenues they are getting on Facebook Watch.
OTT video usage is driving the penetration of various Internet connected devices to help view online streams on the larger TV screen.
Major Hollywood studio to trial 'virtual' movie theaters using head-mounted displays.
Network technology vendor Sandvine has found that piracy isn't only hurting network operator profits – each pirated set-top box is also using up 1TB per month in 'phantom bandwidth.'
On-the-Air Thursdays Digital Audio
ARCHIVED | December 7, 2017, 12pm EST
Orange has been one of the leading proponents of SDN and NFV. In this Telco Transformation radio show, Orange's John Isch provides some perspective on his company's NFV/SDN journey.
Special Huawei Video
Huawei Network Transformation Seminar The adoption of virtualization technology and cloud architectures by telecom network operators is now well underway but there is still a long way to go before the transition to an era of Network Functions Cloudification (NFC) is complete. |
|
|
||
Telco Transformation
About Us
Contact Us
Help
Register
Twitter
Facebook
RSS
Copyright © 2024 Light Reading, part of Informa Tech, a division of Informa PLC. All rights reserved. Privacy Policy | Cookie Policy | Terms of Use in partnership with
|