Netflix: The Paradox of Choice

Conventional wisdom says that we are better off with more options. But a jam experiment done in a grocery store at Menlo Park, California by Sheena Iyengar and team in 2000 proved otherwise. When loaded with more choices people often get overloaded with options creating a situation called Analysis Paralysis and often end up not taking any decision. While there is contention regarding the different domains in which choice of number of options matters, the experiment’s findings are more relevant in this information age than ever.

Sounds Familiar? It’s the story of my life and many Netflix subscribers. According to recent statistics, an average American watching Netflix spends 40 minutes(90 minutes if you take only active subscriptions) viewing movies and TV shows. More and more titles, movies, foreign movies get added to Netflix everyday(still GOT will be added only in 3015) making us spend more time browsing latest, search by genre, search by keyword, related movies and we are lost in browsing only to realize we have spent so much time on browsing and exhausted pick up some random chick flick or binge watch some popular show(which might still be the best decision but not the solution you were expecting)

Earlier we had VCR stores, and then came CD/DVD stores like Blockbuster. There was an informational overload even in the stores but it was limited by the size of the store and the information available with the user. Fellini was known only to a few and even though this generation knows about Kubrick and have greater access to his work, the number of people who actually watched his work will not be very high compared to the disc era.

Netflix transitioned itself from a DVD delivery company to a Video Streaming company and with original programming (though in nascent stage), the options have become more than ever.


There are 2 ways to access content in Netflix:

1) Search by keyword

2) Browsing through Category

This design works when the size of library is small or if you know exactly the kind of product you are going to search like Amazon. It is very important to be really quick in terms of transactions, payments, loading pages in e-commerce kind of setting and Amazon does a good job to provide that user experience. Netflix does an equally good job in this area and also in streaming the content but a search similar to Amazon does not solve the Content Search problem of the users. It’s time to redesign the search.

TV Programming — Limited Options

We are quick in transitioning everything from disk to cloud, from cable to internet but we don’t take the learning from those models which were hugely successful earlier. TV’s basically had programming what you call in today’s terms Curated/Editorial Playlists. Programs were designed, slotted and channels made sure that there was variety, popular as well as experimentation content. And there was not just one channel. They made sure that Independence Day was shown on July 4th instead of typing in Google “Movies to watch on July 4th” to give an internet discussion recommending Independence Day in the middle of so much noise. We abandoned the working signals to do something by our own because we believed more options will help us make better choices.

Machine Learning and Recommendations

Machine Learning can help streaming sites model user watching behavior and then can cluster similar people together and can provide recommendations based on your patterns/group patterns, etc. Sky is the limit for recommendations when you have data at your disposal. Studies indicate Netflix re-engineered Hollywood. It can determine the combination of actors/directors/plots which could be the next big recipe for success. After main stream TV Channels shunned Unbreakable Kimmy Schmidt, Netflix picked it up and it is a runaway success. Netflix might have known the demographic it might appeal to and the new customer acquisitions it can bring.

But these recommendations make our filters even deeper; the suggestions take us into the road already taken. I’m not complaining when it makes my decision fast like watch HIMYM because you watched Friends but eventually I will end up watching romcoms about 6 people living in NYC.

Netflix should become my Siri instead and eventually Scarlett Johansson.

User Profile and Internal Social Network

I am sure Netflix collects information about a user by their viewing history, social media profile ( if you connect FB) but it definitely lacks the information which Google has and I am pleasantly surprised by the nice card it suggests me every day about which movie to watch based on the searches I made. Though Netflix has introduced suggest movies to friends, I am not sure about the success of the product at least by the minimum buzz it failed to generate(or all my friends on fb ignoring me). May be it is the privacy concerns or for whatever unknown sociological problem of millennials, the possibility of what our friends watch will be a good set of recommendations to start with instead of relying on comments by unknown people( I am always surprised by the quality of Netflix comments, great work here).

I always search for a movie on Netflix even though knowing the probability of knowing that movie will not be on Netflix and get zero results which are usually followed by that movie being magically picked up by Netflix(Murphy again) and I am kept in dark about that development. Instead of those messages which get piled up in my mailbox, it would be great if I get a message saying

Hey Santosh, I know that you wanted to watch Night Crawler yesterday, in case you were not crazy to watch it on popcorn time or even more crazier to buy it on Amazon, we covered that for you — come back let’s watch together.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.