July 29, 2019
If you’re new to machine learning, it can be difficult to get your head around how it might apply to your business. While nearly every industry collects, processes and uses information in some capacity, how does this translate to understanding data with machine learning?
In our new guide, we discuss how to find competitive advantage in the machine learning era. One example we touch on is Netflix. We challenge you to find anyone who doesn’t know what this business does!
Below, we explain how understanding data with machine learning is helping Netflix to delight global audiences and dominate the media streaming market.
Many of us suffer from short attention spans, and this can impact our interactions with many brands. Worryingly, Netflix found that customers tend to give up if they spend more than 90 seconds searching for a movie. This means in less than two minutes, a viewer can lose interest, stop engaging and potentially browse a competing service. The stakes are high.
However, Netflix utilises machine learning to improve search results and minimise audience frustrations. By analysing searches and generating tailored recommendations, Netflix reportedly saves US$1 billion a year in potential lost revenue.
Netflix also analyses past viewing data with machine learning to maintain streaming quality, predicting bandwidth usage and reacting to peaks in service demand to keep viewers happy.
Netflix has come to be known as a tastemaker within the global film and TV industry. This is no big surprise, considering that 80% of viewing choices made by Netflix customers stem from the platform’s personalized recommendations.
Streaming services are beginning to better understand data with machine learning, and are able to recommend shows that customers actually want to watch, while also unpicking stereotypes and assumptions they may have previously relied on.
Historically, audience demographics were determined by age, gender and location. However, today we know that there’s much more to it. Many adults love Twilight, women love true crime documentaries and UK viewers love Scandi-dramas, for instance.
With huge audiences around the world, and the ability to analyse all this data with machine learning, Netflix has a heightened understanding of these trends. As a result, its subscriber base is divided into 1,300 taste communities, which are based solely on past viewing behavior rather than outdated demographics.
Far from sorting films into broad categories of comedy, drama, horror or romance, Netflix also organises movies into an incredible 76,000 categories.
Netflix claimed there were 33 million different versions of the streaming platform back in 2013. At that moment in time, Netflix had 33 million subscribers in total.
Not only does Netflix personalize the content it suggests, it also personalizes the way in which it suggests it. In fact, by analysing data with machine learning, Netflix is able to adapt the entire user experience to each individual subscriber. This includes the rows selected for the homepage, the titles they show and the cover image for each title.
This personalization is driven by what Netflix refers to as “consumer science”. Each element aims to highlight how a film or show is relevant to the viewer’s interests.
The New York Times once claimed: “Netflix is commissioning original content because it knows what people want before they do.”
All the knowledge gathered from understanding data with machine learning has served Netflix well. By gaining a deep understanding of consumer interests and trends, not only can the brand best serve them in the present - it can plan ahead to delight them in the future.
Last year, Netflix directed 85% of all new spending towards original programming. By taking control of content production, they don’t need to compete for the rights to popular shows, and are able to create the kind of content they know their audience is looking for.
Netflix spent six years gathering enough consumer data to create their first original content, the globally successful ‘House of Cards’. Since then, this formula has helped to achieve success rates of 80% compared to 30%-40% success rates of traditional TV shows.
After reading this blog, you’d be forgiven for thinking that Netflix bases all its decisions on a wealth of consumer data with machine learning algorithms at play. However, the reality is actually very different.
While consumer behavior is meticulously monitored, the results do not dictate creative direction. Even in a data-obsessed company like Netflix, humans are still in control of the final decision-making process, which is “70% gut and 30% data.”
This corroborates our own beliefs on understanding data with machine learning. Technology won’t replace humans, rather it is simply a cape to be worn by your team of data heroes.