It’s hard to image Netflix as the DVD mailer it was 20 years ago, it’s business built on old-fashioned “snail mail.” With is evolution over the years, it has become one of the largest streaming video platforms in the world, using the latest in technology innovation to bring solutions to its users.
With over 100 million users logging in each month, Netflix faces challenges on many fronts: changing mobile technology, internet speeds and data limitations, user tastes, content formats, and new video streaming competitors launching constantly, just to name a few.
How does it anticipate the next round of tech advances to beat out its competitors? In this three-part series, we’ll dig into the areas Netflix is forecasting and how they use the Artificial Intelligence to continue their growth.
Part 1: Overcoming Buffering Frustration
One of the biggest problems online streaming services must overcome is preventing videos from buffering — the start-stop, wait-while-it-loads frustration. And believe it or not, this is one of the main focuses for Netflix’s Artificial Intelligence technology — keeping the customer happy with a smooth, consistently-streaming product.
This is especially difficult as the company seeks to expand to countries with limited internet access. As Netflix grows its viewer base, and its reach includes Korea, India, and countries in Africa, the issue that arises over again is how to provide seamless viewing experience with slow, unreliable internet. But, even in metropolitan locations with higher bandwidth (like Chicago), buffering is still a consistent concern.
Imagine this scenario for a high bandwidth location:
It’s a rainy night and the user has finally selected a video from millions of videos available to them. They are five minutes into viewing and the buffer circle appears and remains in place for quite some time.
After frustrating attempts to refresh, the video remains frozen or stop-and-go.
Eventually the user will move on to a different selection or log off for a different alternative for video entertainment. If Netflix isn’t known for delivering a stable product, the users can’t trust they will be able to use the service when they most want it.
Let’s look at how Netflix’s AI solves the buffering issue for their users:
To avoid painful situations, Netflix has poured in hours of time, research, and funding into their AI bitrate algorithms, developing an approach called “per-title encoding.” These identify complex (versus simplified) video frames and assigns a customized viewing compression.
The AI algorithms compare the quantity of bits per frame paired with individual users’ internet speed. The AI is trained to analyze the effectiveness of a scene’s compression rate against another one after hundreds of viewer studies and research. This allows Netflix to provide a better, more stable viewing experience using less bandwidth.
An example of a high bitrate scene would be a battle scene from “The Lord of the Rings” battle scene with explosives and CGI warriors. Another could be a motorcycle and car chase in one of the “Mission Impossible” movies. Because there are so many moving parts and fast-acting three-dimensional elements, more bitrate is used versus scenes of dialogue. These scenes require enhanced assessments from Netflix artificial intelligence to better distribute bandwidth.
Netflix has also assigned lower bitrates to less popular programs or genres when compared to other movies that have higher viewer ratings. This helps reallocate the optimal resources to the highest viewed videos, impacting the highest amount of viewers.