Pioneered by engineers at Google and Microsoft for search engine marketing, scientists at the startup Globys want to infuse machine leaning and automation into real-time A/B testing for campaigns running in other advertising media, both online and offline.
From email to display advertising and even online marketing, marketers run tests to measure the impact of certain elements in the campaign. Machine learning -- which can automate functions like discovery, testing and optimization -- gives marketers the ability to test thousands of permutations simultaneously to determine the efficiency of each targeted promotion or message. So why would marketers want to manually determine the winner between "A" and "B"?
Globys executives call the technology "self-optimization personalization." It uses the brand's content management system and behavioral analytics from first-party purchase histories to optimize targeted advertisements for desktop and mobile. And the beauty of mobile, said Glenn Pingul, VP of scientific marketing strategies at Amplero, "it's not too often you hear, 'here, honey, take my phone for the weekend.'"
The core premise of A/B testing works, but machine learning will take advertising into the future to measure the long-term effect of one message versus another. Four years in the making, Globys officially rolled out Amplero earlier this year. Olly Downs, chief scientist of the Amplero platform, admits that the technology tends to work best when targeting a brand's existing customers, rather than using it for branding or acquisition marketing. The method works better with first-party data -- individual actionable data. The technology requires the brand to know the customer through the life cycle of the relationship, but Downs expects that will change in time.
Influence requires repeat interactions. Retail, hospitality, gaming, travel and other vertical markets that have multiple methods to reach consumers, from in-store to online and email, where consumers tend to become loyal followers seem to leverage the technology best.
"Think about an online game," Downs said. "I download the app and try it for free. I enjoy it. During the game it adapts to my liking making it more interesting to engage me in the game. I become an ongoing game player and someone the brand can up-sell me on additional values."
If the game player frequents the mobile apps of Web less often, the technology can determine when and why, targeting new offerings that bring him back.