How can marketers use AI to create better customer experiences?

19 Nov 2018

Image: © Worawut/

While many marketing departments are aware of the potential benefits of digital tools such as machine learning, actually implementing them can be an overwhelming task.

Digital transformation is on the minds of many enterprises, from C-level executives to security teams and marketing departments. A key element of this transformation is the use of machine learning (ML) to create artificial intelligence (AI) systems that can automate repetitive tasks.

Although there is plenty of buzz around the potential business gains to be made, particularly in marketing, it can be difficult to create a manageable implementation strategy. spoke to senior director of product marketing at DataRobot, Colin Priest, about how marketing departments can seize the opportunities these technologies offer.

Different strengths, different weaknesses

Priest said it’s helpful to look at the strengths and weaknesses of humans and computer systems. By examining them this way, the best marketing strategy can benefit from the help of both. Using machines, tasks such as mathematics, data manipulation, parallel processing at scale, and repetitive processes can become much simpler.

While digital systems have the edge on humans in some respects, Priest said engagement and communication are on our side. “Humans are better at listening, and selling and evangelising ideas.”

As well as this, general knowledge and application of context to situations is another uniquely human advantage. Human creativity also trumps digital systems. “Not only are humans better at music and art but if a customer has a non-standard problem, humans are much better at creatively solving those problems.”

Empathy as an advantage

Perhaps most crucially, the human ability to be empathetic and understand ethical systems gives them the edge over machines. “Your customers are human. They are inherently social and emotional. Research shows that when banks pushed customers to use ATMs, customer satisfaction dropped.

“Customers need to interact with humans. This is particularly true for any issue that is emotional.” Priest pointed to incidents whereby customers could not solve a problem using an FAQ page alone, but a human staff member could provide a novel, creative solution. “Recently, I flew overseas and when I landed, I switched on my cell phone. Within a minute, my phone told me that I had run up more than $100 in international roaming fees.

“I tried to switch on roaming via the telco’s app but it wouldn’t let me. I needed a human to help me fix this problem.”

When it comes to manual processes, this is where ML can shine, said Priest. “ They [customers] don’t want the paperwork, red tape or other annoying processes. These tasks should be invisible.”

Four golden rules

Priest boiled it down to four simple rules in business:

  • If a task is transactional, then automate it.
  • If a task has well-defined, predictable outcomes and needs to be done at scale, then automate it via AI.
  • If a task is emotional or social, then let your customers see a human face as soon as possible.
  • If a non-standard problem arises, empower your AI to pass it to a human to solve, and enable your customers to talk to human staff members for help.

Individualised marketing is an emerging trend, as segmentation becomes less of a best-practice strategy. This is an area where automation can lend a helping hand. “Customers don’t want to be arbitrarily grouped with other customers just because of their age, gender etc. Your competitors have already realised that and treat their customers as individuals.

“They look at all of their customers’ characteristics, from their demographics to their past purchases and behaviour, and they use AI to link those characteristics to predict future outcomes.”

An AI can learn by example, spotting patterns that customers eschewed in the past and using this to predict the probability that another customer will churn. Priest noted: “It is now best practice that the AI will tell you the reasons why it made that prediction. These actionable insights enable you to retain valued customers.”

Predictive analytics

Along with managing and reducing churn, predicting future actions of customers is also a possibility. “Once again, it is best practice that the AI tells you why it ranks some customers as more valuable than others, whether it be because of their characteristics or their past behaviours.”

Combining these customer rankings with churn predictions can help marketing teams focus on retaining the most valuable customers or users.

Priest also noted the importance of explanations for customers around decisions made by machines, particularly in a more regulated environment. “Customers have the right to an explanation of an algorithmic decision. You can’t use black-box AIs for marketing any more. This is driving businesses to upgrade their AIs to the latest generation, which provide human-friendly explanations for their decisions.”

AI can also improve customer perception of communications from the business. As opposed to using technology to power ‘next-best offer’, it can be leveraged further towards choosing the right communication at the right time: ‘next-best action’. “These communications are not always about selling a product. Frequently, they are about engaging the customer via useful information and/or experiences. They also adapt to customer behaviour.”

This comes back to the individualised response customers expect, Priest concluded. “Some customers want to quickly get to the purchase decision, while others need much more nurturing before you offer a product or service to them.

“AIs can learn from the past and individualise the customer nurturing experience.”

Ellen Tannam was a journalist with Silicon Republic, covering all manner of business and tech subjects