Online recommendation systems have been touted to change the way we interact with technology, work, media, emails and so much more by creating a personalized system of communication and recommendations we enjoy receiving. However, we often use online recommendation systems in our everyday lives without ever realising that we are using them. At Boomtrain, we often have potential customers who want to personalize their marketing mix and make it more engaging, but can’t understand the potential of recommendation systems to do this. To make this easier, we put together a list of the brands we all love today and how they use online recommendation systems to personalize their content and drive success.
It isn’t often that a movie and television show streaming website gets a mass following to the degree that a whole phrase in the world of coupling is named after it. Then again, not every streaming website is Netflix. ‘Netflix and chill’ did as much for couples as it did for Netflix, which took the United States by storm in 2013 – moving its DVD rental business to an online streaming entity that created its own TV shows.
In January of 2016, the Netflix team did the unthinkable and launched in over 190 countries, making the odds of success in every country fairly low. Yet, Netflix has consistently done one thing right ever since they became a streaming service, and that is investing in online recommendation systems. According to a Netflix tech blog, the company uses a combination of machine learning, personalized ranking and page generation to give us the recommendations that get us hooked, show after show.
Okay, not everyone loves facebook anymore. Most of us love to hate it, and wish we spent less time on it. However, the one thing facebook is really good at, is getting an audience hooked. Ever since the facebook newsfeed came into existence and made us endless newsfeed-scrolling zombies, users have wondered what online recommendation system is used to create our personalized news feeds.
Image: Social Media Today
Addictive as it is, there is little information about what exactly facebook’s algorithm does to create the news feed. Buffer does a good job of tracking facebook’s updates and alerting audiences about what is changing. Factors such as how long you spend reading an article, to how often you look at what your friends post all go into this mysterious algorithm to make your news feed more engaging and personalized
That awesome new webseries you found yesterday? You can thank YouTube for showing it to you. Or should you be thanking the online recommendation systems that YouTube uses? According to a Marketing Land piece written in 2011, YouTube measures factors like click through rate (CTR), long CTR (only counting clicks that led to watches of a substantial fraction of the video), session length, time until ﬁrst long watch, and recommendation coverage (the fraction of logged in users with recommendations.)
After watching this cute kid tell his mother her pregnancy was “exasperating” I just had to see him on Ellen.
Personalization and Image: YouTube
YouTube uses Amazon’s recommendation algorithm to make predictions – maybe that explains the awesomeness – and recommendations account for about 60% of all video clicks from the home page. The power of personalization, folks!
How do you get your morning earworm out of your head? I usually turn to Soundcloud. The Berlin based music listening app has always kept the playlist going once your song of choice ended, but never has been as precise as today.
Soundcloud’s ‘suggested tracks’ or ‘Discover’ feature now lets you find similar music based on music you have listened to in the past, and the results are actually pretty great. Guess what they use to create this? Machine learning algorithm based online recommendation systems! For someone who has been using soundcloud a while, the amount of material you can discover that is personalized for your taste is endless, and of course, awesome.
If you’ve ever shopped on Amazon, then you’ve experienced their age-old recommendation engine that somehow manages to still get it right everytime. Amazon created their ‘item-to-item collaborative filtering’ algorithm close to a decade ago, and it has served them, and all their customers well ever since. Personalization is what drove Amazon’s item-to-item filtering, and their email recommendations, both of which are a huge success.
This year, Amazon announced that it was releasing the API to its online recommendation systems, particularly the AI engine ‘DSSTNE’ (Destiny) to help organisations use machine learning and harness the power of datasets even sparser than the one’s Google’s TensorFlow is capable of handling. Will Amazon’s release democratize the recommendation capabilities of the rest of the ecommerce world? We’ll have to wait and see.