What Is Machine Learning’s Impact on Marketing Personalization?

Marketers by now know machine learning has made an impact on marketing personalization efforts. Machine learning includes personalization use cases in the areas of lookalike modeling, churn analysis, cross-sell/upsell analysis and many others, according to a Forrester report last year on machine learning’s future. “Machine learning powers targeted ads, personalized content, song recommendations, and Amazon Alexa,” Forrester lead researchers Mike Gualtieri, Brandon Purcell and Kjell Carlsson wrote in the report. “The internet giants are all-in on machine learning because it is the only way for them to satisfy hundreds of millions of users with smart, personalized services. … (Machine learning) can plumb the depths of data to contextually understand your customers and help you design scalable, hyper-personalized customer experiences.”

So we know the “what” about machine learning’s impact on personalization. What are some other important considerations for marketers to know regarding machine learning’s impact on personalization?

Machine Learning and Personalization Vendor Landscape

First, it’s good to have a sense on what the marketing technology vendors are offering for machine learning capabilities. Gartner researchers Jennifer Polk, Martha Mathers and Jason McNellis provided a snapshot of some of the most significant personalization engine vendors in their Magic Quadrant published last July. “Machine learning is proliferating in personalization engines,” McNellis, senior director analyst at Gartner for Marketers, said in an interview with CMSWire. “In our 2019 Magic Quadrant for Personalization Engines, vendors had as many as 14 prebuilt predictive machine learning models. Presence of machine learning is becoming ubiquitous, but the quality is not evenly distributed, so marketers need to know how to evaluate the potential of prebuilt AI for their specific use cases.”

Here are some of the Gartner findings on vendors in that personalization engines vendor report:

• Acquia Lift: stand-alone personalization engine that delivers personalized web content based on machine learning content recommendations
• Adobe Target: applies Artificial Intelligence (Adobe Sensei) and machine learning for scoring, segmentation and personalization 
• Certona: participates in the Google Cloud Partner Program, providing access to Google’s machine learning
• Dynamic Yield: Users can upload their own algorithms to enable native machine learning, outlier detection and removal
• Evergage (recently acquired by Salesforce): Uses machine learning to build unified customer profiles supported by diverse predictive scores
• Oracle: Includes Oracle Adaptive Intelligence Apps machine learning to define segments. 

Related Article: How Bringing Machine Learning Into Marketing Improves Business Results

Machine Learning Growth in CDPs and Beyond

Gartner also found Customer Data Platforms (CDPs) includes some built-in personalization engine machine learning capabilities through standardized data feeds. And as for the personalization engines themselves, broader input data “has the potential to improve machine learning core to many of these platforms.”

“Legacy vendors are increasing their investments, such as Salesforce’s continued investment in Einstein and their recent acquisition of Evergage,” McNellis told CMSWire. “Newer platforms have often included machine learning since their inception but continue to expand their functionality and simplify their workflows so marketers so they can take better advantage of the capabilities.”

Machine Learning Focus Falls Outside of Marketing Activation

Marketers should recognize that leveraging machine learning for personalization efforts will almost certainly include third-party vendor help, according to Jonathan Treiber, CEO of RevTrax. Many machine learning initiatives, he said, are typically outside of marketing activation, in the realm of data analytics and business intelligence. Marketers, he added, are finding some minor success incorporating machine learning in their marketing. “Most markers are relying on third-party vendors to enable them with ML-powered solutions, versus looking internally within their own organization for data science expertise and hoping for IT roadmap prioritization,” Treiber said.

Related Article: Cutting Through AI Marketing Hype: It’s About Machine Learning

Machine Learning Connects Individual Touchpoints

Bernard May, CEO of National Positions, told CMSWire one of the ways that machine learning is impacting the personalization processes in marketing is the individual touchpoints made with a customer. “Beyond just adding names or business names to automated emails, marketing automation systems can learn which contact methods clients interact with regularly and deliver messaging via those preferred channels — email or, SMS, or chat,” he said. “Part of personalization is not just connecting with a potential client, but doing so in the way they prefer.”

The same is true for communication timelines, May added. “Marketing automation systems can understand not just if a client is being responsive but also how quickly they respond,” he said. “Some clients may take two hours to respond on average, others may take two weeks. So, machine learning can be used to plan communications with clients based on a timeframe that work best for them.”

Advanced Analytics Meets Homegrown Personalization Efforts

With machine learning already woven into many personalization tools, can marketers successfully use advanced analytics for homegrown personalization efforts? Yes, according to Gartner for Marketers’ McNellis. This can work for slower-velocity use cases, such as email marketing, where message personalization doesn’t need to be done in real-time, though it can be improved by it, according to McNellis. Other use cases such as web or mobile app personalization, or messaging that relies on real-time triggers, often benefit from real-time incorporation of current session data and any available contextual variables. “This type of personalization,” McNellis said, “generally requires a platform such as a personalization engine which has built-in data platform, content management, customer analytics and decisioning.”

Related Article: True Marketing Personalization Takes Talent, Technology and Empathy

The Power of Data

The data that powers machine learning is just as important as the algorithmic approach, McNellis said. This is likely, he said, a big factor in Acquia’s recent acquisition of CDP provider AgilOne. “We’re seeing vendors improve the speed and flexibility of storing and updating customer profiles to keep up with marketer demand,” McNellis said. 

Vendors are developing solutions that move beyond time of day and device type to include inventory levels near your location or hovering behavior online. McNellis is also seeing a trend around incorporating unstructured data, both text and images, to drive better content and product recommendations. In Gartner’s Marketing Technology Survey 2019, generative content creation ranked second for its potential positive impact on marketing. “Incorporating and modeling unstructured data,” McNellis said, “will be key to making that a reality.”