#Ad – this post contains an advertisement
Streaming platform Netflix has seen an explosion in subscriber numbers thanks to lockdown. Chances are, you have a subscription and have binge-watched your way through a fair few box sets.
But how does Netflix keep you hooked and looking for that next binge-worthy show? They don’t leave it down to chance. Alongside its marketing and word of mouth campaigns, Netflix also relies heavily on AI to run its industry-leading recommendation systems.
Recommendations engine
The Netflix recommendation AI uses a combination of content-based recommendation and lookalike audiences in order to present you with tailored recommendations. Now, that might sound like complicated mumbo jumbo, but it’s actually relatively simple The algorithm looks at what you’ve been watching, or ate showing an interest in, and makes recommendations based on your user history, genre preference etc.
This approach will only work for so long until you completely run out of a particular genre. In this instance, Netflix looks at other users, with similar tastes to you and uses their viewing behaviour to predict which films and TV shows they think you will like.
It’s working too. Over 80% of users choose movies based on Netflix’s recommendations and they’re refining the system all the time. It’s no surprise that the big streaming players are using AI in this way, machine learning is already being used heavily in our professional and private lives to some extent. Companies like Industrial Vision Systems, Quantum, NetGuru are producing advanced systems in healthcare, marketing and data science. Use the Uber App? That’s based on a type of AI too.
Beyond its recommendation engine, Netflix also uses machine learning to influence our viewing behaviour in other ways.
Profile-based recommendations
A while ago, Netflix introduced the option of creating profiles for other users within your household. Super helpful, especially to those of us with kids, or wildly different viewing tastes than our partners. Instead of your recommendations being clogged up with Paw Patrol, Netflix needed a way to be able to accurately recommend what to watch, rather than making the best guess based on an apparent viewing history that is a combination of many people’s viewing habits.
Artwork personalization
It may surprise you to learn that the artwork that accompanies a movie or show on Netflix isn’t the same for everyone. Netflix knows that you sometimes do judge a book by its cover, so they set about changing that cover to be more appealing to you.
The algorithm will take into account things like your viewing history, age, gender and location before presenting you with the artwork it believes will have the highest chance of getting you to watch it.
Adjusting optimal streaming quality
The quality of your streaming is very important. A buffering, low-quality stream will have you switching off in no time. That’s not all down to your home broadband connection. With so many viewers, offering the highest quality streaming to everyone is almost impossible, that’s why during lockdown there was a furore about them reducing the quality for many subscribers in Europe in order to keep up with demand.
Netflix can now, with a high degree of accuracy, predict demand for particular titles and make sure they are positioned in strategically planned server positions in advance.
So there you have it. How Netflix is using data science and machine learning to bring you recommendations and entice you to watch. As technology becomes more advanced, it will be interesting to see how they use it in the future.
No Comments