Cognitive Decision Making With AI: Using Customer Signals

COLLIDE Data + AI Conference 2023Cognitive Decision Making with AIWatch the full session · 21 min

At COLLIDE 2023 in Atlanta, Sol Madadha, SVP of Data Analytics & Governance at Regions, argued that businesses win when they combine the many signals customers give off into one picture of what each person wants. In a roughly 21-minute talk on "cognitive decision making with AI," Madadha said he manages data analytics and governance for Regions Bank's wealth management business.

His case, as presented at the time: customer preference data has become abundant, AI can make sense of it, and the payoff depends on clean data, a clear use case, a return on investment and governance.

Why do customer preferences matter for AI?

Madadha opened by asking ChatGPT for a phrase on customer preferences. He said its answer, that knowing preferences is "the key to deliver products and services that Delight, engage and retain customers," could not be improved on.

He traced his own interest to a B2B e-commerce project he worked on in Atlanta about 23 years earlier. The task was to find three key preferences for each business, price, quantity and time, and score every available offer against them. That was cutting edge then, he said, but the volume of signals about each customer today is far larger, and much of it is available in public marketplaces.

Is AI the next big technology disruption?

Madadha described a sequence of disruptions that changed business applications: the internet, e-commerce, smartphones, social media, big data, cloud computing and blockchain. AI, he said, "is going to be the biggest thing you'll ever see."

He cited Gartner forecasts of growth over the following three years, including 30% of new applications using AI to drive personalized, adaptive user interfaces by 2026, up from 5% at the time. Those figures are the speaker's reading of the forecasts. He added that he thought such projections tend to be conservative, recalling a 1999 forecast of Amazon revenue that he said was later doubled by reality.

What AI use cases did the talk highlight?

Madadha mixed consumer and enterprise examples. He pointed to autonomous driving, citing how a Tesla identifies objects with an onboard computer. He also described AI-assisted editing in Photoshop and Adobe's Firefly, including virtual staging of rooms with furniture.

In financial services he named trading platforms, fraud detection and customer service as areas where AI would have a large effect. He also mentioned summarizing a whole book from a URL in seconds with ChatGPT and listed predictive modeling and personalization among the main uses.

How do multiple signals lead to a decision?

The core of the talk was a personal story. Madadha said he test drove a Tesla Model X and, as the shop was closing, received an offer to keep the car overnight. He reasoned the salesperson had his driver's license, proof of insurance, a credit report showing he could afford the car, and a customer profile showing his web browsing on the company's site.

"All this information put together" let the salesperson make the offer, he said, and his interest in buying rose "from 25% to 75%." He called this combining of signals "cognitive senses," comparable to a person combining sight, hearing and touch.

He extended the idea to other settings:

• Airport check-in. A startup he spoke with was working on recognizing a traveler through face, phone and wearable signals to speed up check-in.
• Location-based marketing. Offers that appear when a customer is near a store, an idea he said he studied in a 2005 master's program but that lacked the technology then.
• Voice of the customer. Feedback, net promoter scores and referrals as additional inputs.
• Car dealerships. A salesperson who already knows what the customer wants and has a car waiting.

What does it take to adopt this in the enterprise?

Madadha listed the prerequisites. "Everything starts with a very good data, clean data," he said, followed by use cases chosen to improve customer experience rather than "for sake of implementing AI," and an investment that delivers a decent return.

Governance was the other big item. He called AI a "Beast" to control and said that with great power comes great responsibility. Technically, he pointed to data integration, analysis that makes sense of the data, real-time processing, scale, and the ability to build AI and machine learning models, which he described as a stack of layers.

Where to learn more

The AI Enterprise Conference 2027 takes place Tuesday, August 31, 2027, at Pier Sixty in New York City. COLLIDE 2027, billed as the largest AI conference in the Southern U.S., is set for Wednesday, October 13, 2027, at Sandy Springs Performing Arts Center in Atlanta. See /events/collide/ and /events/ai-enterprise-conference/. Registration for 2027 events is coming soon.

Frequently asked questions

What did Madadha mean by cognitive decision making?

He used it to describe combining signals about a customer, such as identity, credit, insurance and browsing behavior, the way a person combines senses, to reach a decision.

What does enterprise AI adoption require?

Madadha listed clean data, use cases that improve customer experience, a decent return on investment and governance.

Which industries did he say AI would affect?

He cited financial services, including trading platforms, fraud detection and customer service, along with examples in automotive and creative software.

How did he describe past technology shifts?

He listed the internet, e-commerce, smartphones, social media, big data, cloud and blockchain before calling AI the next big disruption.

This article is based on remarks by Sol Madadha, SVP of Data Analytics & Governance, Regions at COLLIDE Data + AI Conference 2023 on October 3, 2023. More coverage: COLLIDE AI Conference.

Filed under: AI Use Cases & Applications, AI Strategy & ROI