Before AI can make sense of all this, the information must be based on coherent concepts and described the same way everywhere it’s stored. If the quality of the data is poor—if auto coverage is described as a “policy” in some places and as a “product” in others, for example—an insurer won’t realize the full potential of its AI tools and the customer likely won’t benefit, either.

To ensure that AI works well, companies need to improve their data management, keep information up to date, and make sure definitions are consistent. “You can’t automate a mess,” says Earley. “You can’t automate what you don’t understand.”

From travel to entertainment, home improvement to banking, companies are realizing how crucial it is to personalize the online experience.

Starbucks, for instance, is able to draw an increasingly detailed picture of its customers with every order placed through its mobile app. Then, with AI, it can deliver offers unique to every customer. It’s a strategy that helped Starbucks generate 52% of its total sales in 2021 from its app.

“It’s really about understanding the consumer beyond a demographic,” says Leala Crawford, who heads Deloitte’s data science analytics and personalization team.

Early data analytics generally provided a one-dimensional view of the consumer, such as age or gender, Crawford says. AI, in contrast, can capture and analyze a wider range of data, and use it to recommend actions based on a fuller picture of the customer. One Deloitte client, a global travel and hospitality company, uses AI to assist service agents helping customers complete their bookings. The data gathered during that interaction—such as travel itinerary or planned outings—is then passed on to the service agents who can then recommend alternative vacation packages.

The payoff for companies is quantifiable, Crawford says. Deloitte research shows that companies who accurately anticipate customer needs generally see 40% improvement in loyalty, retention, and revenue.

In the insurance industry, processing claims can be a drawn-out and maddening process. AI chatbots and conversational tools can improve the claims experience for both policyholders and adjusters. Not surprisingly, insurance companies have been at the forefront of chatbot adoption.

With AI, a chatbot can cross-reference claims, check for possible fraud, and process a payment within minutes. It also makes it easier to purchase insurance. For instance, someone pricing life insurance online can enter the relevant data—their birthday, whether they smoke or have a family history of health problems—and the AI system instantly evaluates risk and provides a quote, satisfying customers’ need for speed.

“For customers, knowing the ballpark of how much insurance is going to cost is going to help you think concretely about how much coverage you can afford,” says Laura McKiernan Boylan, head of underwriting solutions at Haven Technologies, a subsidiary of MassMutual.

Other industries also are successfully using AI-chatbots to streamline customer workflows. Banks are reducing case volumes and response times, while large telecom companies are cutting down on “status requests” by using conversational AI to let customers know about interruptions in service.

AI-powered conversational tools let customer-service representatives focus more on what they do best. “That customer service representative is able to do relationship-building,” says Nicole Chiala, senior manager at ServiceNow’s Workflow Design Studio. “That’s what people do really well, and machines can help us with the other pieces.”