How Netflix uses big data selectively to stream supreme


Posted By Pinar Dost ⋅ March 4, 2019

There aren’t many companies that have seen their products and services become part of international pop culture as rapidly or decisively as Netflix. The media streaming giant, which started out as a mail-order DVD rental service, has over 139 million subscribers across 190 countries at last count, and now boasts a huge collection of exclusive content unavailable anywhere else. Much like ‘Google’ is now a colloquialism for searching online regardless of search engine choice, ‘watching Netflix’ is starting to enter the vernacular as a way to describe binge-streaming on any platform.

Not only is Netflix one of the world’s biggest internet companies, it’s also officially the world’s most data-hungry application. According to an independent bandwidth company, Netflix consumes 15% of all global net traffic – an astonishing feat for a service that’s only been streaming internationally since 2012.

Of course, for all the data Netflix users devour, they’re also generating a vast amount too. Here’s how Netflix uses (and doesn’t use) data analytics to keep its subscribers coming back for more.

Personalised user experience

Netflix itself has famously said that there are millions of different versions of the platform ; a different one for every user. It means this literally; sign in on someone else’s account and you’re likely to see a completely different set of content on the home screen as you would do on your own. This isn’t accidental – Netflix tracks every viewing decision you make and bases the type of content it promotes to you on what it thinks you’ll enjoy – not only in terms of the programmes it recommends, but even down to how they appear to you.

Here’s an example. Many Netflix users will be recommended Stranger Things, it’s a Netflix Original and one of its most popular shows. But if you’re a viewer who usually opts for sci-fi or horror-based titles, you’re much more likely to see a landing card (the image that appears in the menu) of the four main characters in front of a lurid post-apocalyptic-looking sky, than an image of the show’s romantic leads, Jonathan and Nancy. Netflix would likely save this landing card for viewers who tend to watch romantic titles and rom-coms, on the bet that it’d be more likely to catch their eye.

In fact, Netflix is so confident in its ability to predict which content users will engage with based on viewing history and other metrics, it doesn’t even include traditional consumer profiling demographics, like age and gender, in its recommendation system. Instead, it uses the wealth of preferential data it holds about each account holder to hyper-personalise user experience on an individual level.

Licencing the right content

Netflix owes much of its success to the fact that it’s the only platform where users can watch some of the most popular content around, but how does it know what’s going to be a hit with viewers?

As well as overseeing production of original programming, Netflix buys the right to stream many externally-produced shows and films, but it doesn’t choose these based on guesswork. Each investment is made after careful data analysis, such as of similar titles and genres, and other titles that feature the same director or cast. By examining what percentage of viewers watched all available episodes of a similar series, for example, or how many watched similar shows and shows featuring the same leading actors, Netflix can decide whether or not a title is worth licencing.

It even takes pirating into account; after all, heavy illegal streaming of a show is a good indication that there are plenty of people who want to watch it.

Prioritising behavioural data

Netflix got rid of user reviews in 2018, realising that any number of factors could lead a user to leave a bad review of a show or film – rather than simply not enjoying a title. In fact, bad reviews were being left on titles reviewers hadn’t even watched, which not only ran the risk of turning users away from content they themselves might enjoy, but skewed Netflix’s own data about user sentiment.

What’s more, Netflix also came to the conclusion that there’s far more to learn from the way users actually interact with the service, rather than what they say about it. Much like it can afford to ignore demographic data, Netflix ranks behavioural data over publicly-visible user reviews, which can be highly influenced by social and cultural bias. More valuable is an analysis of how long users watched a particular title, common turn off points and the number of repeat viewings, for example.

Using consumer data intelligently

We’ve only scratched the surface of how Netflix analyses user data, the level of detail it goes to is truly impressive. Yet, just as important as the data it chooses to drill down into, is the data it chooses not to take into account – knowing there are more effective ways to find out about its users than what might seem like ‘go-to’ metrics.

Understanding which data is and isn’t useful should be a key consideration for any business in an ever changing landscape. At Quant, we can work with you to narrow down the noise to the specific consumer information essential to your business’s unique objectives. This brings you closer to your customers and identifies ways to expand your market reach. Simply get in touch to find out more.

Share Button

Categories: Big Data, Customer Data, Insight, Seize the Data


Related Articles: