Machine Learning is ablaze just now, and marketers are noticing that many of the companies and applications at the leading edge are aimed squarely at the heart of their trade.
Google’s DeepMind may be grabbing headlines for winning at Go, but it’s in areas like market research, personalization and recommendations where machine learning is having real impact, and not in ways you might expect. Consider AI start-up Emotient, which Apple recently acquired. Emotient makes software that recognizes the emotional content of facial expressions, which marketers use to assess people’s reactions to advertising and product features. Such approaches are rapidly replacing traditional market research techniques as they help marketers gain new insights into the emotions that drive spending.
But smart machines aren’t just analyzing things: they’re also starting to create content that sells. Consider Persado, which markets a “persuasion automation solution” that produces machine-generated marketing language tailored to the top-performing emotions associated with a marketer’s target customer segments. Or Resonate, another company that uses machine learning and predictive modeling to target ads based on motivational emotions, serving brands and political advocacy groups. Clearly we’re seeing a trend that seeks to use data and smart technology to unlock the power of emotion to sell.
Confronted with this incursion of algorithms into the sanctum of human emotions, how will a marketer react? Amusement? Disbelief? Enthusiasm? Shock? Dread? Whatever the initial response, we can identify four broad areas of consensus:
- This is not going away
- Underestimating the potential of data and algorithms to model emotion can be a crucial mistake
- However adept machines get at analyzing the data of emotions, empathy remains exclusively human
- Underestimating the importance of empathy in marketing can also be a crucial mistake
To put some structure around these ideas, my colleague Jake Sorofman and I have developed the Intelligent Brand Framework (subscription required). In it, we examine examples of how brands are balancing the best practices of data-centric marketing with the human-centric approaches in the areas of automation, observation, inspiration, and engagement. This week we’ll look at more examples of how data and machine learning are breaking boundaries and, at the same time, freeing marketers to focus their attention on the things algorithms can’t do. The combination will be far more powerful than the sum. In the meantime, don’t let your emotions cloud your judgment about machine learning: it’s powerful stuff, and you’ll need three things to get ready for it:
- Lots of data for training sets
- Some skilled staff or partners
- A culture of experimentation
And maybe one more: the judgment to know when to trust your head and when to trust your heart.