12 min read
Machine Learning is (or should be) a core component of any marketing program now, especially in digital marketing campaigns. The following insightful quote by Dan Olley (EVP of Product Development and CTO at Elsevier) sums up the urgency and criticality of the situation: "If CIOs invested in machine learning three years ago, they would have wasted their money. But if they wait another three years, they will never catch up." I believe that this statement also applies to CMOs. Here are some of the top machine learning trends in marketing that we see developing:
Machine Learning-based personalization (SegOne = Segment of One Marketing) is hotter than ever, especially when marketers select context-specific content to be presented to an individual consumer. Content-based marketing is one the highest-ROI approaches to marketing because it delivers something surprising, interesting, relevant, and useful to the consumer. This is especially true if machine learning (ML) is used to match the right content to the right user at the right time in the right context (i.e., Context-based Marketing). This means that ML will be applied to customer interaction data in order to find and exploit patterns in different types of user interactions that occur at different times and locations: purchase histories, emails, call center interactions, social media, website searches, previous marketing campaigns, and even location data and/or “emotion data” from wearable sensors while the customer is shopping.
The integration and blending of multiple diverse data sources is leading to an age of hyper-personalization. For example, one customer noticed that, after searching on her home PC for a specific kitchen gadget for her home, a popup ad appeared on her husband’s computer suggesting that he buy this item.
The benefit of delivering hyper-personalized content is that it really moves the needle in the marketing funnel, by driving the customer from awareness, interest, and conversion to the higher realm of loyalty and advocacy, ultimately having the customer become your vocal fan in the marketplace.
There is greater use of social channels to learn about and engage with consumers, including real-time ML-based sentiment analysis, leading to just-in-time marketing experiences at the moment when a consumer mentions your product online in a manner that indicates that they could use some customer service (i.e., social customer care). In one case, a customer gave a very positive review on social media of his hotel check-in experience – the next morning he found a gift box of his favorite beers waiting outside his hotel room door with a thank you note from the hotel manager (all of which was made possible by mining the customer’s social discourse).
Consequently, sentiment analysis is no longer being used just to get the "large sample" pulse of overall population sentiment after some marketing campaign or event, but sentiment analysis is used for one-on-one real-time interactions with an individual customer. In order to make this feasible and affordable, process automation will be required to drive these interactions. Customer experience automation is AI-powered, enabled by ML.
The application of AI should be broad enough to include both human-assisted machine intelligence and machine-assisted human intelligence – i.e., Augmented Intelligence (not Artificial Intelligence). The benefit of this approach is that ML can help to automate redundant processes (via machines), improve the quality of customer interactions (via humans), and make recommendations that are timely and responsive to a customer’s immediate needs (via Augmented Intelligence).
The number one hottest trend (where ML-based marketing is moving) is behavioral analytics, including both predictive and prescriptive analytics modeling. Understanding what motivates a consumer, how a consumer signals their intentions and interests, what turns off a consumer, what conditions induce a consumer to make a decision, etc. are all part of behavioral sciences. Predictive marketing models help to answer the question "what is the consumer likely to do next?", while prescriptive marketing models help to answer the question "how can our marketing investment achieve a better, optimal outcome in this situation, hopefully producing a win-win for the consumer and for the business?"
For example, a mobile telecommunications provider was able to determine when any of its customers are about to leave the country (based on location data, e.g., the customer is in the international departure terminal of an airport). Consequently, knowing that international travel often causes mobile customers some concerns about their mobile access while outside the country, the company used this information to offer (profitably) a discounted international mobile roaming call package to such customers “just-in-time” prior to their departure.
To apply these concepts successfully within marketing requires access to comprehensive (360 view) data sources, ML model-building capabilities, and an agile analytics environment that is dynamically responsive to changing conditions.
Conversational chatbots are definitely hot in the customer engagement space! Many such applications are currently in customer care call center operations, providing AI-assisted (ML-powered) personalized care to individual consumers. You could say that these customer interactions are mainly reactive (i.e., responding to a customer's requests). But, we should be on the lookout for more proactive and prescriptive customer interactions with chatbots – this is a specific example of the prescriptive marketing that we mentioned above, only this time it is audio-based.
For example, there are several startups that are mastering the art and science of ML-based Natural Language Generation. So, it is not a big leap for us to imagine a marketer using customer data to generate a personalized video ad to a unique customer that appears on one of their social media timelines. Maybe the image stream in the video is not specifically personalized (at least, not initially), but the audio track can definitely be personalized. ML makes possible that personalized Natural Language Generation from customer data sources.
Another hot trend in marketing is Agile Analytics, which was referenced earlier. ML is first and foremost a set of algorithms that learn to detect and recognize patterns in data, and to learn from experience when making decisions or taking actions based on those patterns – i.e., the ML process learns (from both successes and failures, including failed models and failed classifications) how to improve over time. The faster and more agile that the learning takes place, then the more impressive will be the customer engagement. Consequently, marketing will no longer focus on static customer experience (measured in hindsight) but will focus on interactive customer engagement: ML-powered foresight, insight, and next-best interaction with the customer.
A great example of this is the Customer Decision Hub developed by Pegasystems. This AI-powered customer engagement system incorporates and exploits customer data, interactions, and feedback in real-time using ML to provide an agile next-best action recommendation to a company’s customer-facing staff. In one case, a client of Pegasystems software says that all of their employees are now "front office" – they are each responsible for delivering exceptional customer service through best-in-class customer engagement, thereby completing the transition from high-latency passive (reactive) marketing to agile prescriptive (proactive) marketing!
In the world of online marketing, it is important to remember also the power of influencer marketing. ML (including social network analysis and social graph mining) is now helping marketers identify the key influencers within their market domain. Consequently, by engaging directly with those influential individuals to help spread their message, a marketer can expect to see the impact of their campaigns amplified to even wider, yet more targeted audiences than could otherwise be reached.
Influencers are no longer just popular entertainment or sports figures. Influencers exist in both broad and very specific niche market domains. Using ML to identify those influential marketing partners is powerful social data science that should definitely be considered as a high-ROI component in all online campaigns.
Journey Sciences (a methodology that is being championed by the folks at ClickFox) is right up there with the hottest emerging trends in ML-driven digital marketing. The application of graph models, link analysis, and graphical causal networks is now making big strides forward in predictive and prescriptive marketing. The customer’s journey is really the customer’s story. Exploring and exploiting that story through journey analytics is golden. The power of storytelling (using stories to engage with others at the right place and time with the right message) is as old as human society itself. The use of ML to model and learn from all customer touchpoints and interactions will therefore take a big step forward when marketers use the most natural representation of the customer experience, which is the customer’s journey. We can expect to see ML-powered journey sciences becoming an increasingly popular approach in many human analytics use cases: customer, patient, employee, or any digital user.
Finally, implicit in many of the trends mentioned above is one more: the growing presence and use of ubiquitous sensors – the Internet of Things (IoT). The IoT is more than just smart thermostats or smart refrigerators in your home, or sensors on machines and engines in industrial environments. The IoT represents a means to collect rich customer data as well as volumes of contextual data (time, location, events, weather, social sentiment, systems, etc.) associated with the customer’s current environment. The streaming data from IoT sensors everywhere will then provide a powerfully rich view of the customer.
Applying ML to build predictive and prescriptive models of a specific customer’s needs, behaviors, intents, and motivators is driving marketing innovation and delivering game-changing marketing insights. Make your marketing move with machine learning now – if you wait much longer, you may never catch up.
To enable and empower this, you need a data platform with a broad set of capabilities: with streaming and batch analytics, structured and unstructured data capabilities, scalable components, fast access to multiple diverse data sources, fast processing, graph analytics, machine learning, and event-driven microservices, while being endowed with best-in-class open source data analytics tools and services. Look no further – MapR has the capabilities and the platform that you need. From the Converged Data Platform, to the implementation of the full Spark stack, to an agile microservices-based architecture, to free training courses, and more – you won’t find a more complete, forward-looking, robust solution on the market than MapR’s. Don’t just take my word for it, but see what others are saying:
Stay ahead of the bleeding edge...get the best of Big Data in your inbox.