Machine Learning Masterclass #2: How is Machine Learning being used in digital?
Welcome to our Machine Learning Masterclass series. Here we will be talking about the importance of Machine Learning and how this connects with the world of digital marketing.
In the first instalment of our Machine Learning Masterclass series, we introduced the topic by explaining exactly how Machine Learning can be understood. In this next article, we explore ways that Machine Learning is already being used in digital and offer some interesting industry applications to reflect upon.
Google and RankBrain
When people think of Machine Learning and Google, they often consider Google Cloud and the services Google offers for businesses looking to delve into Machine Learning. Yet, don’t forget that Google itself is using Machine Learning through its artificial intelligence system, RankBrain. Part of Google’s overall algorithm, RankBrain concentrates on refining the plethora of search queries which Google encounters. In detail, every day Google tackles three billion searches, among them complex and unseen longtail queries. How does RankBrain make sense of these? It identifies patterns in seemingly unconnected complex queries and learns how to understand future complex searches and whether they relate to certain topics and themes. It can then associate these groups of queries with the results that it believes are most relevant to users. Understanding how Google itself is using Machine Learning is therefore crucial for SEOs.
TripAdvisor and text analysis
An obvious example of how Machine Learning is used for text analysis and interpretation is through chatbots in customer service. Perhaps, a more surprising use of Machine Learning for a similar purpose is how TripAdvisor is employing AI to assess how useful each review left on the site is. As we know, central to any Machine Learning dilemma is a huge data set that cannot be processed by humans alone. TripAdvisor receives nearly one trillion reviews every week – a human team could not realistically determine if every one of these reviews is helpful or not. So, TripAdvisor uses a text classifier to “read” reviews, create a helpfulness score, and decide whether the review should be automatically rejected, automatically published, or queued for a human to moderate it further.
Amazon, Netflix and recommendations
Platforms such as Netflix and Amazon use Machine Learning in their algorithms for making recommendations and suggestions to users. Again, they aim to make our consumer experience as relevant as possible and improve engagement. How is this achieved? By recognising the input, finding relevant searches, predicting which results are most useful and returning a ranked output.
Uber and time prediction
One of the best examples of how Machine Learning can become integral to all areas of a business is by delving into the popular taxi app ‘Uber’. In 2015, Danny Lange became the head of machine learning at Uber and highlights: “We have found over time that Machine Learning does add value in areas where people initially didn’t think of Machine Learning as an option.” One of the key ways that the machine ‘learns’, is by collecting ‘big data’ from millions of trips to be able to accurately estimate when a car will get to you, at any time of the day, whatever day of the week it is.
Snapchat, Facebook and image recognition
We’re also coming across Machine Learning in image and speech recognition more and more frequently. For example, platforms are now able to recognise human faces in images – this is how Snapchat uses ‘Faceswap’ and Facebook asks you ‘Who’s in this photo?’ when tagging. What’s more, your iPhone creates a ‘People’ album of faces it recognises as the same person! It can get a little confused when someone is wearing sunglasses, however you can only expect this technology to continue to improve. In terms of speech recognition, Siri and now Alexa are the most obvious examples, yet translation platforms are another good use of speech recognition.
While these developments are great for the digital world, one of the most exciting things about Machine Learning is its propensity to dramatically improve everyday life for people all over the world. For example, ‘DuLight’ by Baidu is an early prototype designed for the visually impaired which uses image recognition to recognise what is actually in front of a person and describe it to them.
Target and advert targeting
A key challenge to other areas of modern digital marketing is the struggle to personalise messages to ensure a greater success rate of every consumer engaging, spending and becoming loyal. In this sense, digital marketing is benefitting from Machine Learning through targeting adverts for specific recipients at certain times. In turn, this improves consumer experience as we’re getting adverts which are more relevant and useful. One of the most famous examples of Machine Learning in advert targeting is when the US retailer Target found out a teen was pregnant before her family did! Target’s statisticians have worked out that when a certain 25 products are bought together it indicates that the woman is most likely pregnant. It can then be worked out when exactly to send relevant coupons for baby-related products at various stages of the pregnancy term, with the hope that targeting future mothers at this crucial phase in life will gain long-term, loyal customers.
This article has explored exciting ways that Machine Learning is already being used in search and across the wider digital marketing industry. But our Machine Learning Masterclass isn’t finished here! Keep your eyes peeled for our future posts in this series, exploring the tools used in Machine Learning and lots more.
If you can’t wait for the next instalment, why not chat to one of Tamar’s technical SEO experts and find out how Machine Learning could be used in your business.