Is Marketing the Clearest ROI Path for Artificial Intelligence?

When most companies think about artificial intelligence (AI)—and its many sub-categories, such as machine learning, natural language processing (NLP), or cognitive computing—they think about applications such as IBM’s Watson technology being used to diagnose patients. What they often miss, according to Fern Halper, VP and senior research director of advanced analytics at TDWI, is that “marketing is often one of the first departments in an organization to use advanced analytics.”

In a recent webinar, Halper discussed the current potential and future pathways for marketing analytics with David Stodder, senior research director of business intelligence at TDWI, and Wilson Raj, the global director of customer intelligence with SAS. They all agreed that among all the different business departments, applying AI to marketing could provide the most immediate, recognizable return on investment .

Raj was quick to redefine AI:

“I think that the main way to think about AI is to define artificial intelligence … in terms of the tasks that humans would do, rather than how humans think.”

AI doesn’t need to understand why a customer might make a particular decision, because a business doesn’t need to know that information. All it needs is for AI to use past actions to make logical predictions about what customers will do in the future.

Using artificial intelligence in marketing

For example, a company might start a campaign to up-sell customers on a new feature that requires an additional monthly premium. Machine learning can be used to quickly analyze every customer’s interaction with the company, their response to previous campaigns, and their path to a transaction. By pulling this data together, predictive analytics can find the portion of customers who are likely to respond positively. And computers can do this work much faster than people—often in real-time—which helps deliver the ROI the company is looking for.

Halper was quick to point out that NLP can be used to better understand the “voice of the customer.” She says, “Customers are also stating their opinions and preferences in many ways. They’re ending emails to many organizations. They’re tweeting about many products online. They’re filling in online reviews. They’re calling customer service agents to complain. All of this involves natural language processing.”

By using NLP to extract semantics from text, companies can use the results to respond to negative comments on social media, or even route a customer’s call depending on how much they spend, whether or not they’re currently happy with the service, or based on their particular vertical’s needs.

Raj pointed to Swisscom, a major Swiss telecommunications company, which was struggling under the weight of customer service requests and complaints. A staff of five people were handling 20 percent of queries on a daily basis. After implementing machine learning technology to process those incoming requests more intelligently, a staff of four people were able to handle 100 percent of queries.

Halper and Stodder also spoke about next-best offers, marketing optimization, and the move from web analytics to “digital intelligence,” which is being able to react to web interactions in real-time, rather than via a monthly update after the fact. Raj added spam filtering, calculating lifetime value, recommendation engines, and real-time ad placements to the list of arenas where marketing analytics could provide value.

Raj insists that companies need to begin by either collecting data, or beginning to understand the data they already have. From there, discovering what marketing arenas have the most potential for their particular business is a priority. A successful deployment of marketing analytics is not just pushing out interactions to the consumer, but also having the internal capabilities to visualize and use decision management.

Between developing models to predict repeat customers with 76 percent accuracy, or reducing customer churn, marketing analytics might just be the next big move for AI. Even if it’s not about replicating human thought, AI can help meet goals every company has : maximize revenue and keep customers coming back.

Related:

Tools and tactics: machine learning

How NLP-based AI systems work