What Should Machine Learning Actually Learn About Your Customers?

The buzzwords for the future are undoubtedly “machine learning” and “automation”. But talking about technology and actually using it to drive your business and reach new customers are two different things. In many cases it comes down to personalization and having the data to create strong customer experiences.

What actually is machine learning? Essentially, it’s the part of artificial intelligence that allows computer models to recognize patterns in existing data so that it can learn what to do or predict what will happen in the future. Just like how humans can absorb information and use it to formulate their observations or predictions, machine learning can do it without bias and by quickly consuming much more data than any human could ever sort through.


You Knew Netflix Used Machine Learning To Tailor Preferences, Did You Know It Earned $1Billion A Year

In the customer world, machine learning helps absorb all of the existing data about customers and shrink it into small nuggets that companies can use to create accurate and personalized experiences. For example, much of Netflix’s algorithm is based on machine learning—with each show a customer watches or rates, the system can learn their preferences and make more accurate recommendations in the future. Part of the reason customers love Netflix is because they can find shows they might not know about otherwise. It’s a nuanced system to find connections and patterns in a user’s preferences, but when done right it can spit out recommendations of shows customers love. And it’s working—Netflix’s machine learning recommendations brings in an estimated $1 billion a year alone.

There are a lot of things machine learning can find out–many directions machine learning can take the company. But what should machine learning actually learn about your customers and where should it focus its efforts? Consider these three areas:

Purchasing Patterns Through Machine Learning At Target and Amazon

Perhaps one of the easiest things for machine learning to track is what customers purchase and when they purchase it. Machine learning can easily track if a customer buys laundry detergent every three months or shops for swimsuits at the beginning of every summer. Taking it a step further, machine learning can then predict what a customer will buy next and make recommendations. Target’s machine learning program can recognize when a female customer starts purchasing things like hand sanitizer, vitamins, and unscented lotion that she is likely pregnant. From there, the store can make subtle recommendations for pregnancy and baby-related products.

Machine learning can track where each customer is in the purchase journey and customize its recommendations. Amazon has one of the best recommendations engines on the internet, with a staggering 55% of sales driven by machine learning recommendations. Understanding customer purchase patterns helps increase sales and also makes predicting inventory demand much easier, especially for seasonal and trend-based items.

Behavioral Patterns Tells Marketers How To Engage

Machine learning can learn how long it takes a customer to decide to make a purchase. Some customers are impulse buyers and make purchases right away, while other people sit around with items in their shopping carts or continuously return to the site to look at the product before they make the decision to buy it. Machine learning can find the patterns of how long it takes each customer to pull the trigger, and then brands can personalize their reminder emails, promotions, and timing to engage with customers and move them along the sales journey without seeming too pushy. A customer that needs time to make a decision will get different messaging than a customer that makes quick decisions and has already moved on to their next purchase.

Machine learning can also know how customers like to communicate. It can bring together information from all the various communications methods including chat, text, phone, and email to synthesize customer data and find patterns. If a customer prefers to connect with a brand in a certain way, the company can match their preferences for a seamless, personalized experience.

Personal Preferences At Pinterest Via Machine Learning

Machine learning can detect basic demographics about customers and website visitors to know their age, location, household income, marital status, and more. This can be used to create targeted ad campaigns. If the algorithm determines that females in their 30s are the most likely to be on the fence about making a purchase, the brand can target its messaging to reach those 30 something year old customers.

Another example is Pinterest Mining personal preferences is at the heart of Pinterest’s business model. The company uses machine learning to find out what users are interested in and then predict other bins and products they might enjoy. For example, if the site knows that a married woman in her 40s has a history of looking for pins about domestic travel locations, Pinterest can recommend other travel-related pins, and businesses can connect with customers in its target demographic.

Machine learning is a powerful tool that can help companies in a variety of ways, especially when it comes to connecting and learning more about their customers.

Blake Morgan is a customer experience futurist, author of More Is More, and keynote speaker. Sign up for her weekly newsletter here. Go farther and create knock your socks-off customer experiences in your organization by enrolling in her new Customer Experience School.

Share this post