How humans find good stuff on Netflix
Designing an engine for personalized content recommendations is hard. Really hard.
So hard, in fact, that searching for something to watch on Netflix has long been a trope of the internets.
In a perfect world, the journey from “wanna watch” to “watching” would look something like this:
That’d be nice. But reality looks more like this:
Recommendation engines (henceforth reco-engine) are big business. They contribute so much to user experience, it’s funny how UX designers don’t really know much about them. In many ways, their performance is as critical to UX as having internet. As the old saying goes,
Nail the algorithm, conquer the world
Calling Netflix an extremely successful company is an understatement. As of January 2018, they’re worth over $130 billion. With it’s continuous growth, it’ll surely soon be worth more than Disney ($155b) or Comcast ($169b).
So they must be doing something right. The three legged stool approach to their reco-engine seems to cover all the bases, or at least the bases they should reasonably cover.
“The three legs of this stool would be Netflix members; taggers who understand everything about the content; and our machine learning algorithms that take all of the data and put things together”
—Todd Yellin, VP of Product Innovation at Netflix
The endless search for good stuff doesn’t seem to be stopping anyone from using the product. And perhaps the current state is just the balance we need.
Perfect personalization requires perfectly personal data. But how does protecting personal data and privacy impact producing a perfectly personalized presentation of programming? If Netflix presented precisely what we pine for, it’d probably perturb people.
As a species we’ve created work arounds—desire paths—to more efficiently and less painfully get from “wanna watch” to “happily watching”.
So I did what any overly-intrusive designer would do, and talked to a bunch of people about how they find good content on Netflix. And came up with a four archetypes to explore the role of our humanness in finding what to watch.
Turns out the reco-engine plays only a tiny part.
The feels determine the reels
This archetype is the emotional type. Her preferences are driven by her current mood, the mood she wants, or the collective mood of the small group she watches TV with, whether human or canine.
This mood analysis brings her to a genre. Her internal dialogue usually goes like this: “Been a long week, I’m tired, don’t want to think, need a comedy, a stupid one, not so stupid it’s bad, but just the right amount of stupidity to be cheesy yet engaging.”
The result is usually not good. Lots of scrolling, cross referencing with google searches, and in the end, settling on something that doesn’t look terrible.
Here’s how Miss Moody finds something to watch.
The unwritten social dynamics of bros in the wild
This fella loves a night out with the boys. And who doesn’t? Because in the end, who knows you better than your bros? 😅
An important thing that came out of this is how the dynamics of a gathering of friends facilitates decision making. Usually, more cooks in the kitchen is a recipe for disaster. But when there is an unspoken trust between the cooks, maybe things change.
The bros, in this case, understand each other and what they want. They have varied tastes, but they can decide what to watch easily because they trust their friends, and are willing to compromise.
Each individual’s definition of the perfect movie is put aside for the greater good. Here’s a simplified flow of how the bromance plays out.
Fam reacts only
I trust, therefore I watch
This goes a level deeper than the bromance. This archetype is about that special person whom you trust with all entertainment recommendations.
In this day and age, it’s a lot to give someone that kind of power over you. One day over coffee they say “hey have you seen Lost?” Next thing you know you’re starting season 6, questioning all your choices in life.
Committing to a new series has become a major life choice, requiring the sacrifice of health and wealth.
But you trust them because they know you intimately, and you know them. There’s transparency in your relationship. A transparency that you don’t get with reco-engines.
I think Drake said it best,
You know what I like, Oh yes, oh yeah, Oh yes, oh yeah, Oh yes, oh yeah
You know what I’m sipping, I’ll teach you how to mix it
But you’re the only one, ’cause I don’t trust these b******
I don’t, I don’t trust these b******
Here’s what the flow looks like.
Rainy Day Romp
Weather gotcha down
Some people, more than others, are severely impacted by the weather. A windy, sunny, or snowy day can significantly impact how they spend their time.
Rain is her kryptonite. Nothing does it like a rainy day. It has that magical power of giving you permission to do nothing and procrastinate on all your responsibilities. Time freezes.
For this archetype, rain also induces a mood. It’s hard for her to put it into words, and it’s more of a feeling of what’s right. After all, the heart wants what it wants.
Interestingly, this archetype was the most open to browsing Netflix, and exploring new options based on the reco-engine. Maybe because they are in a relaxed state of mind, without a sense of urgency, and all the cares of life—like the rain outside—have fallen to the ground.
Here’s what it looks like.