“Machine learning” may not be a term you’ve heard before. But if your first impulse is to ask Alexa what it is, you’re already using one application of it.
Machine learning is the practice of getting computers to learn autonomously, through the accumulation of data input. The more input, the more accurate the algorithms become. In other words, machine learning is getting computers to learn the way people do – through experience. It’s a form of artificial intelligence.
Machine learning has a huge number of applications. Voice and facial recognition, analysis of healthcare data for research and diagnostics, detection of online fraud, even lunch ordering—a service called Forkable uses algorithms to predict what everyone in your office wants to eat, then delivers it. And if you’ve ever been served an ad on Facebook and thought, “How does Amazon know I need garbage bags?”, then you’ve experienced the marketing side of machine learning.
How marketers are using machine learning
-
Personalization. By collecting data on customers’ behaviors, machine learning gets to “know” them, and can predict the kind of products or messages they’re likely to respond to. So many companies use this kind of personalization that consumers have come to expect it. In the future, marketing messages that are clearly NOT personalized will stand out (and not in a good way).
Elements of personalization can include names, geographic location, driving and shopping habits, and more. The British company Bidooh, which sells digital outdoor advertising (such as large posters in shopping malls), uses facial recognition to choose which ad it serves to passersby.
-
Email segmentation and optimization. Instead of laboriously dividing email lists into preference groups, companies can use machine learning to help target more accurately according to behavioral patterns. Machine learning can help generate myriad combinations of subject lines and messages, then optimize for the ones that perform best; it can also optimize timing and frequency of when emails are sent. Some email services even use machine learning to generate the content of emails. One app, called Phrasee, claims that it “writes better-performing marketing copy than humans.”
As a human writer who needs my paycheck, I’m not sure I like the sound of that. But some say it’s the way of the future. According to a 2017 whitepaper by IDC Research, “Programmatic advertising has automated many of the transactional elements of the advertising supply chain, and IDC believes that the next bastion is the creative content and copy process.”
-
Forecasting. With machine learning, you don’t have to be a visionary to predict future demand or anticipate customer behaviors. For instance, real-time analysis of data can help pinpoint customers you may be about to lose, then deliver them a special message or offer to bring them back. It can also help determine the optimal media mix to reach your target audience in the months ahead.
Can machine learning go too far?
Machine learning is a great tool. But beware of the “creep” factor. In a 2017 survey, 36% of consumers who chose not to buy a smart speaker (like Amazon Echo or Google Home) cited concerns about privacy as their reason. In October 2017, a glitch in the Google Home Mini caused the device to secretly record users without their knowledge, which did little to change skeptics’ attitudes.
Likewise, marketing that oversteps its bounds by being a little too personalized can trigger the creep factor. Spotify pushed the boundaries in 2016 when it created posters that called out the actual listening behaviors of some of its customers. One headline read, “Dear person who listened to the ‘Forever Alone’ playlist for 4 hours on Valentine’s Day, you OK?” While this is a clever and amusing ad, if you were the person in question, it would be undeniably disturbing. Many people find the sort of ads that call out their current location (“We see you’re shopping at Target!) unsettling as well.
Marketing technologies are developing so rapidly, learning to use them correctly is often a case of trial and error. Sometimes it’s hard to know where the line should be drawn until you’ve already crossed it. But the possibilities are exciting.
