Cognitive Marketing: Explained

By Anne Field

How a data-based approach to anticipating consumer behavior will change marketing forever

In January, HBO premiered The Young Pope, a series about a good-looking, hip guy who also is the first pontiff from the United States. To promote the launch, BETC and Havas Cognitive, working on behalf of the client, the French premium cable channel Canal+, tried the usual — a sassy social media campaign.

But the execution wasn’t your run-of-the-mill promotion. Instead, Havas Cognitive used IBM’s Watson artificial intelligence (AI) capabilities to scan Twitter for people discussing anything related to one or more of the seven deadly sins. Then the AI-enabled engine searched 39,000 verses in the Bible to find the most relevant missive and tweeted it back to the appropriate person, with a link to the show’s website.

Ultimately, the system responded to more than 1 million tweets over the course of the campaign, all the while learning more about how to reply and getting better at offering the perfect quote. “It found people talking about touchy subjects and quoted the Bible back to them in a tongue-in-cheek way,” says Marc Maleh, global director at Havas Cognitive. “People loved it.”

The promotion not only won a Cannes Lions Innovation award, plus a good deal of press, it also shined a spotlight for brands on what experts are calling cognitive marketing — an approach that mixes AI, natural language processing, facial and image recognition software, and massive data analysis to offer multiple consumers highly personalized, contextualized content and messaging that would otherwise be impossible to generate. “Marketers can have a true one-to-one relationship with customers,” says Marc Blanchard, global head of experience design at Havas. “You can talk to millions of people the way you typically would be able to talk to one person.”

In other words, the promotion provided a glimpse of what marketing increasingly will look like as more consumers become accustomed to the incorporation of the technology in their everyday lives — and brands and agencies are obliged to follow suit. According to a 2016 study by Demand Metric, 80 percent of marketers say personalized content is more effective than the less-personalized variety. “Understanding this technology is an immediate competitive imperative for all marketers,” says Gerry Murray, research director of marketing and sales technology at IDC.
 
Collecting One’s Thoughts

In fact, one of the biggest benefits of this technology is the ability to gather vast amounts of structured and unstructured information, massage and analyze it, and come up with content that is most relevant to a specific individual. “There’s a ton of information you can get from social feeds about a person’s valued and deep-seated needs, like perseverance and playfulness, that lets you optimize the experience, content, and product,” Blanchard says. That can mean, for example, grabbing information from social media sites, layering it on top of demographic information, and figuring out what an individual might want to buy. Murray provides the following illustration:

“Your analysis might reveal that a customer talks a lot about boating in his social media interactions, so you might send him content about merchandise that would appeal to someone in his demographic who goes boating,” he says. “You learn about hobbies and interests you couldn’t discern from a standard database or even a branded interaction with a customer.” But the real clincher is the capacity to improve: the more the technology is used, the greater its ability to learn and fine-tune its responses.

Still, it’s only in the last year or two that the technology has become widely commercially available and tech companies have begun baking the capabilities into their systems. Most brands will rely on outside services and experts for this technology. “There are certain prerequisites required for this to work,” Murray says. “You’ve got to have good data sets, data management practices, and familiarity with analytics and model building.”
 
Think of the Possibilities

For now, there also are limits to what the technology can do and how much complexity it can handle effectively. For that reason, the most widely used applications tend to be pretty narrow in focus. “The narrower the swim lane you’re in, the more likely you are to be successful,” Murray says.

Take lead-qualification agents. Also called virtual agents, these software programs analyze leads gathered from, say, requests on a website and other sources, and decide whether the individuals were tire-kickers or prospects worth pursuing. According to Murray, because there’s actually not much variability in the content of such interactions, it’s easy to categorize them according to whether they’re promising or not. (When a response falls outside of certain answers, a human supervisor can be notified to step in). “You have 10,000 leads maybe every month, maybe every week, maybe every hour — too many for people to connect with efficiently,” he says. “But virtual agents can do this really quickly.”

Recommendation engines for content or products are another example. According to Joe Stanhope, VP and principal analyst at market research firm Forrester Research, the technology is considerably more sophisticated than suggesting to a consumer who just clicked on an umbrella description that a few pairs of rain boots might be of interest. “[The systems] have a much more nuanced understanding of the customer,” he says.

Case in point: the furniture retailer Room & Board recently used a recommendation engine from digital marketing platform Salesforce Marketing Cloud. According to Meghann York, director of product marketing at Salesforce, Room & Board realized it needed to create a digital experience that reflected how its customers shopped online. The company traditionally had placed a lot of importance on in-store displays, complete with strategically situated complementary items that resulted in significant sales. How, then, to replicate that experience online, but take it a few steps further?

The answer was to use a sophisticated AI-enabled recommendation engine that combined an assortment of information to make suitable suggestions. In addition, marketers also linked to those results in email notifications sent a few hours later. The result was a dramatic increase in revenue from purchases of recommended products.

In some cases, customer segmentation insights lead to answers to important questions brands didn’t even know they should be asking.

Then there are applications used for gathering smarter insights, allowing for more effective customer segmentation — identifying who is more likely, say, to buy a product or click on an email. Salesforce’s Einstein engagement platform, for example, can look at a customer’s past email interactions, purchases, and mobile and web activity, and build models determining who stands a better chance of opening an email, clicking on it, or unsubscribing. “We used to guess at those things,” York says. “This takes all the heavy lifting off the marketer and makes the process more accurate.”

She cites FareCompare, a travel data site, as an example. The company was experiencing a high unsubscribe rate for its weekly email notification. Using Einstein, marketers recently were able to segment their audience into groups of highly engaged customers and unsubscribers. First, FareCompare offered the more engaged folks special deals. The result: a 13 percent increase in revenue. Then, for those in the unsubscribing segment, the company started sending out emails less frequently, with a 66 percent decrease in the unsubscribe rate.

In some cases, such customer segmentation insights lead to answers to important questions brands didn’t even know they should be asking. IDC’s Murray points to another travel site that did a lot of email marketing, with at least 50 percent of its traffic coming from outside the U.S. The company recently ran a weekend airfare offer globally, assuming that travel habits would be pretty much the same everywhere. But an AI engine’s email analysis of click-throughs and other behavior revealed that patterns varied by region. Travelers in Europe preferred driving to nearby countries and were considerably less responsive than people in Latin America or the U.S., who were traversing longer distances.

A Forward-Thinking Technology

Some marketers predict the next really big leap will be in the area of what Benjamin Lord, global strategist at Kinetic, an interactive agency that’s part of WPP, calls “artificial emotions” —technology able to assess in the moment what products or content customers are likely to want. As an example, Lord and his colleagues are looking at how best to use artificial emotions (AE) to engage affluent Chinese tourists interested in buying luxury items at key points of their journey. With the right technology, a marketer could assess travelers’ tastes, behavior patterns, and travel plans based on insights from WeChat, the most popular social network in China. Then the brand could combine all that data with location-specific data and provide customized messaging about products of interest available where the individuals are at that moment. If they are by the Champs-Élysées in Paris, for example, then the AE/AI-enabled technology could push just the right product from the right store. “It will harness many different variables to trigger the perfect message,” Lord says.

Marketers warn that, for such technology to be accepted by consumers, customers need to see a real benefit. “We’re aware of the creepiness factor,” says Havas’ Blanchard. “There has to be a value exchange: ‘You’re going to give me a premium experience and, in exchange, I will allow you access to a set of data.'” As AI-enabled cognitive marketing becomes more prevalent, it’s a value proposition marketers ignore at their peril.