Skip to content
 

The Surprising Secrets Behind How Netflix Recommends Shows You Watch

Spoiler alert: You may not like what you learn

The Netflix logo is displayed on a smartphone in front a television screen that's on the streaming service's home page

Chesnot/Getty Images

Partly thanks to 2020 pandemic binge-watching, Netflix gained new members approximately equivalent to the population of 15 U.S. states, nabbing a record-breaking total of 209 million customers. But they all have the same question: How the heck do I find something to watch on Netflix?

Instead of trying to figure it out ourselves or typing in a movie or show title, 80 percent of us rely on Netflix's recommendations — brought to you by 2,500 engineers armed with “multi-armed bandit algorithms” and “probabilistic graphical models” that live to please you by tracking what you like. They please you well enough that 93 percent of Netflix's original shows have gotten renewed for a second season, while Stone Age-style broadcast TV kills about two-thirds of the new shows it bet on.

Here are some fascinating fast facts about Netflix's recommendations to help you find what you want. (Spoiler alert: They don't really “know” you.)


AARP Membership — $12 for your first year when you sign up for Automatic Renewal

Join today and get instant access to discounts, programs, services, and the information you need to benefit every area of your life. 


1. Netflix manipulates the thumbnail artwork of the browsing page to try to catch you.

"If you don't capture a member's attention within 90 seconds, that member will likely lose interest,” writes Netflix's Nick Nelson. “Users spent an average of 1.8 seconds considering each title they were presented,” he adds.

Think that browsing folks are spending that 1.8 seconds reading the title and thinking about the show or film? Nope. While on Netflix, 82 percent of our focus is on the thumbnail artwork for each movie or show. “We were surprised by how much impact an image had,” writes Nelson, “and how little time we had to capture their interest.”

So they started trying to please you with an image chosen to match what all the data they're collecting on you says might appeal to your particular interests. They discovered that more than three faces per image drives us away, and that those faces must punch our emotional buttons hard and fast. “Humans are hardwired to respond to faces,” writes Nelson. “Faces with complex emotions outperform stoic or benign expressions.” For action films or kids’ entertainment, villains outperform while nice faces finish last. But it goes beyond that: You're possibly not seeing the same thumbnail your neighbor does. For viewers who like comedies, Good Will Hunting's Netflix page will often depict Robin Williams. Romance fans may see Minnie Driver smooching Matt Damon instead. Comedy aficionados see the kids in Stranger Things dressed as Ghostbusters; crime drama aficionados see an image of a cop in noir-ish fog.

Feel seen? Or spied on?

2. But those recommendations, it turns out, aren't close to perfect.

For one thing, 41 percent of Netflix users watch without paying for it — they're using a friend's or family member's account. That means Netflix is collecting data on several viewers in a group, not each individual — which pleases nobody.

Writer and podcaster Stacia L. Brown was particularly displeased back in 2018 that her Netflix recommendations kept showing her Black faces, like Leonard Ouzts and Blaire Brooks for the film Like Father. “This film stars Kristen Bell/Kelsey Grammer,” she says, “and these actors had maaaaybe 10 cumulative minutes of screen time, 20 lines between them, tops.” In fact, Netflix's algorithm should have no idea of Brown's race, gender, ethnicity or age — or yours, or anybody's. But it sometimes behaves as if it does — “a marketing trick,” Brown tweeted.

Feel seen? Or stereotyped?

3. Some say Netflix recommendations (and perhaps even programming) are becoming increasingly like internet clickbait.

In August, Slate's tech expert Dan Kois skewered Netflix's recommendation images and headlines as “clickbait: luring someone into clicking, and then delivering something other than what the headline made them want.” He mocks his personalized Netflix page for featuring half-clad beauties in sexy poses that he says misrepresent the shows: a bikinied woman for Who Killed Sara?; shirtless hunks for What Lies Below and Lucifer; a woman with an orgasmic-afterglow expression for Sex/Life; a woman wearing nothing but underwear and $20 bills for Heist — a show Kois calls even more generic and clickbaity than a previous Netflix show, Money Heist.

On the other hand, what some call clickbait others call “hits.” Slate (and many others) razzed Netflix's latest show, actually called Clickbait: “It's announcing itself as potentially dishonest and exploitative and daring you to click anyway.” And what was Netflix's most-watched show the week it debuted? You guessed it. As pundit Haley Soen wrote, the “new ‘silly’ Netflix thriller Clickbait is a bit awful, but everyone is obsessed with it.” Critics called it “disastrous,” but Rotten Tomatoes audiences rated it at a sky-high 89 percent.

Feel seen? Or baited?

4. Those rows of shows by genre that Netflix offers you are increasingly specific.

On your personal Netflix page, the algorithm's best guess at what you'll like go in the row at the top, with the most strongly recommended titles at the left — unless you chose Arabic or Hebrew as your language, in which case they go right to left.

It used to be that those genre categories felt pretty generic: comedies, dramas, indies. But have you taken a closer look at your categories lately? They are absolutely baroque-like, AI-generated phrases to make you feel understood or respected. But do they? Here are some of my favorites:

  • Foreign Satanic Stories From the 1980s
  • Post-Apocalyptic Comedies About Friendship
  • Cult Evil Kid Horror Movies
  • Deep Sea Father-and-Son Period Pieces Based on Real Life Set in the Middle East for Kids
  • Assassination Bounty-Hunter Secret Society Dramas Based on Books Set in Europe About Fame for Ages 8 to 10

Feel seen? Or overseen?

How to take control of what Netflix serves you

While Netflix tries to lure you with photos, assumptions and even clickbait, you can fight back! Here's what can help move the algorithm at Netflix to perhaps actually better understand you and offer you shows and movies that you may actually enjoy.

Tell Netflix your real preferences

Don't like what you see? Advise Netflix of your preferences. When you sign up, rate some movies and shows you like — it will change your algorithm results. Go to your Netflix home page, search for a title you like — I like Grace and Frankie — hover over its image, and click “+” to add it to your My List page of favorites, and click the thumbs up or thumbs down icons to feed the algorithm your wishes.

Hunt for genre lists you actually like

One way is to try these sites, http://ogres-crypt.com/public/NetFlix-Streaming-Genres2.html or https://www.whats-on-netflix.com/library/categories/ which list various genres and their Netflix codes. Feel like shopping for Tearjerkers? Simply locate its code from the list of genres (6384), log on to Netflix, and look at the url (universal resource locator, which means website address) at the top. It looks like this: https://www.netflix.com/browse/.

To add a new genre list, add “genre/” plus the Netflix code for the genre that catches your fancy. I almost went for Campy Zombie Movies (1515), but after scrolling through the 45 genres that start with the word “cerebral,” I went with Cerebral Biographical Movies and typed in its code (4518), so that the url read https://www.netflix.com/browse/genre/4518. Presto, up popped Helen Mirren in The Queen (last day to watch on Netflix: Sept. 30), plus a nice list of other titles to scratch that intellectual itch.

And then I had second thoughts. Wouldn't I rather watch Hidden Gem Fight-the-System Movies (Goon, Gasland) or Classic Feel-Good Opposites-Attract Movies (Never on Sunday, White Christmas) or Cool Moustaches (Breaking Bad, Sons of Anarchy, True Grit, Poirot, Super Mario Brothers Super Show)?

But Netflix only helps those who help themselves to its riches — and can bring themselves to make a choice. So take control!

Tim Appelo covers entertainment and is the film and TV critic for AARP. Previously, he was the entertainment editor at Amazon, video critic at Entertainment Weekly, and a critic and writer for The Hollywood Reporter, People, MTV, The Village Voice and LA Weekly.

Join the Discussion

0 %{widget}% | Add Yours

You must be logged in to leave a comment.