25th Apr, 2019

Never Stop (Machine) Learning

After a few months to a year of using AI to optimize your marketing programs, perhaps applying it across channels and using it to inform decisions on marketing content including digital, TV and radio ads, as best practice suggests, it can be tempting to feel you can take it from there. You have the data and insights on the marketing language your various audiences best respond to across channels. AI is part your marketing budget, and it competes with many other priorities. Why continue to spend the money?

Because you want to continually test, learn, personalize, stay relevant, build brand love and make money. Your market, products, the competition, and your customers are all dynamic and changing all the time. None of it sits still, nor should you. How do you keep track of these thousands of variables across multiple channels and millions of customers to continually optimize your creative? Data science and AI is the answer. With data science and AI always on, you’re continually learn about your brand and customers, as well as their personas, what they care about, what works with them and when they change. Remaining static and using the same insights over and over without receiving new ones is dangerous: You will find decay over time and will then be in the position of replicating what you did with the machine on your own. If CMOs and creative directors are Tony Stark, AI is their Iron Man suit.  

A retail company targeting college-aged women may have to deal with more frequent customer changes than a technology company. Each new crop of customers may respond differently to different marketing language and imagery. Perhaps one year the emotion Achievement (Think “You did it! Celebrate another year in the books with fantastic footwear!”) worked well, whereas the next year a more deal-savvy audience was more attracted to the product offer of 50% off. The machine will pick up on this, show your team the insights and supercharge your campaigns accordingly.

And it doesn’t even have to be year to year. Engagement trends seasonally, too. Different emotions might work differently in the summer than during the holiday season, a Persado report based on 300 million message permutations over a three-year period proved. For instance, Encouragement (To motivate and/or inspire one to take an action by explicitly prompting them to do something) was the top performing emotion in marketing messages in April 2018 but fell to the bottom of the rankings in July.

Different elements of a message are larger contributors to engagement on different platforms, as well. Emotional appeal (which words and phrases deliver the strongest appeal to customers) is king on email, contributing to 60% of engagement, whereas Functional language/Calls to Action are most important on web pages, banners and display ads. Formatting (bold, italics and symbols like emojis) is the top reason for engagement on Facebook. AI ensures that, while different elements of a message may be emphasized to increase engagement, the customer journey and brand voice remains consistent. By combining AI with Narrative Intelligence, the ability to build and tell stories using genuine, natural language, consumers will receive a perfectly precise message with their favorite products from one of their favorite brands.

Over time, we see the ads or messages brands run can be personalized into micro-segments, down to maybe four or five people. So, every four or five people will get a different version of a message or an ad. The differences are subtle, but they are significant. I’ve seen subtle changes in the words that brands use to communicate a product or service, call to action, emotions, their narrative can majorly swing engagement. Because every word of the messaging gets more personalized and precise, CMOs see massive uplifts in brand engagement and revenue performance across all these channels. The difference between the best and the worst message can be upwards of 400%. Why risk being on the wrong side of that? And why stay static? Doing so would be short-sighted, and you won’t see the decay coming. Don’t get blindsided.


Subscribe to our newsletter for the latest news & information.