The routine is familiar now. Off from the day, lounging on a couch, ice cream at the ready, the remote or mouse clicks onto the preferred streaming site, perhaps to watch the show everyone is talking about or to be reacquainted with an old favorite.
Streaming television and movie services are so deeply ingrained in the quotidian now, binge-watching is a common weekend activity and there are colloquial dating terms that invoke them — “Netflix and chill.” People log onto Netflix, Hulu, and other streaming services at all times of the day: during their commute; in the morning while getting ready; and when curling up at night.
Yet despite the large number of users and their constant use of streaming services, such sites continue to build completely unique experiences for each and every user, mostly by using artificial intelligence (AI) from complex mathematical equations expected to predict users’ behavior, known as algorithms.
Netflix’s Algorithm Is Anything but Chill
Streaming giant Netflix has managed to weave its algorithms to predict almost all of the interactions on the site. In fact, it very nearly governs almost everything the user sees when they log on to watch a program, including (but not limited to): page construction, genre rows, the trending videos a user may see, the order they see the videos in, and the image shown from the content.
While the team does have human support — a team of 40 freelancers who hand tag the content and more than 800 engineers, according to a 2013 WIRED article — the algorithms do a lot of heavy lifting when it comes to recommendations.
Algorithms are so vital to the site Netflix even reportedly collects data on how users click and browse to better serve their members. Xavier Amatriain told the tech publication, “We know what you played, searched for, or rated, as well as the time, date and device. We even track user interaction such as browsing or scrolling behavior. All that data is fed into several algorithms, each optimized for a different purpose.”
How Netflix’s Algorithms Track and Entice
According to Netflix research written by Carlos Uribe-Gomez and Neil Hunt published by the Association of Computing Machinery in 2015, the streaming site estimates the average user loses interest after 60 to 90 seconds if they have not found a worthwhile title to stream. To mitigate this turnover and increase engagement, the algorithms track several markers to best customize the viewer experience.
“We have data that suggests there is a different viewing behavior depending on the day of the week, the time of day, the device, and sometimes even the location,” Amatriain told WIRED. The algorithm acts accordingly to best foresee and provide for the viewer, accounting for the daily and even hourly experience Netflix imparts on its users.
Such efficiency helps Netflix engage the consumer for longer periods of time and retain its consumer base. This system of the compiled algorithms, VentureBeat reported, helps the company profit “over $1bn a year from retention alone.” The system is described as “core” to the streaming service because it aids in mitigating “abandonment of our service for an alternative entertainment option.”
A Site of Many Algorithms: The Anatomy of Netflix
There are several algorithms Netflix employs that control and personalize the user’s experience. On any given Netflix homepage, there are about 40 genre rows, with about 75 titles, all of which are chosen by algorithms and aggregated by the page generation algorithm.
Genre rows are determined by Personal Video Ranker (PVR). Within the genre rows, the Top N ranker finds the best matches for the user, using short-term trends. Video-to-Video similarity informs the viewer of options they might be interested in based on previously viewed titles.
“By looking at the metadata, you can find all kinds of similarities between shows,” Gomez-Uribe illuminated to WIRED, pointing out similarities in rating and time period that help determine what is recommended. “You can also look at user behavior — browsing, playing, searching.”
Based on what statistic or even image might most interest the viewer, the evidence selection algorithm designates what the viewer may see. Gomez-Uribe told WIRED, “Placement matters. The closer to the first position on a row a title is, the more likely it will get played. The higher up on a page a row is, the more likely it is to generate a play.” Even the in-house search engine is algorithm optimized, generating 20% of viewership.
Hulu and Human Interaction
As compared to streaming giant Netflix, Hulu’s systems rely much less on Al algorithms and more on a human connection. In late 2019, they launched a more enhanced recommendation system, which logged what the user watched and what time, and a more improved search engine. But when it came to trending videos that might interest the user, Hulu still relied on human contact.
“We know that today’s connected consumers expect a deeply personalized experience when they watch TV. At Hulu, we’ve always taken a unique approach to recommendations. It’s a combination of human curation, empowering our viewers, and algorithms that round out the personalized experiences we deliver to Hulu subscribers,” said Jason Wong, director of Product Management.
While algorithms aid the site, human interaction determines far more at Hulu than at their competitor, Netflix. In a statement to The Verge, the streaming site stated: “At Hulu, we believe the best search and discovery experience is built on three key tentpoles — our editors that find and highlight content that is relevant and timely, our recommendation algorithms that work to understand what our viewers like, and our features that enable us to listen to our viewers and give them more control over what they see.”
Not All Feedback Is Good Feedback
But even when all data is collected and compiled, artificial intelligence is just that — artificial. Many users simply don’t rely on their suggested programs at all, only playing what is humanly recommended by friends and family.
Sometimes, users have even remarked the algorithm made life harder, like those who shared an account during a relationship and had to deal with the shared recommendations in the aftermath of a break-up.
Jesus Diaz wrote the following for Fast Company: “I don’t care how efficient the company says the algorithm is — from my personal experience, it doesn’t work. A machine can never fully replace personal taste and exploration based on human interaction.” He continued in the op-ed: “We’re sick of algorithms telling us where to go, who to listen to, and what to watch. Your machine predicts a 98% chance that I would like to watch Frozen. I never will. In fact, I have yet to find an instance of any algorithm surprising me with a smart suggestion.”
Even Gomez-Uribe admitted to WIRED, there are certain limits to algorithmic recommendations. “I watched Tell No One, the French thriller, over a year ago. I’ve been trying to find similar movies. The person on the content team who acquired it said it’s the only one like it in the world.”
Curling up with Artificial Intelligence
While there is likely nothing that will ever replace the buzz of the happening television show everyone is talking about, algorithms certainly influence what people watch while streaming.
The average viewer may not even think about how their favorite streaming site works, but algorithm-driven artificial intelligence is constantly assisting their host site in mastering a delicate swan-like dance: running smoothly on the surface while fiercely paddling to control, create, and sustain a distinctive adventure for each consumer. When someone logs on during their commute or curls up on their couch with the Gilmore Girls, they’re cozying up with their own personal artificial intelligence.