Path to purpose: Ahmed Abbasi explores the interface of AI and humanity

Author: Brendan O'Shaughnessy (ND '93)

Web Istock 1271635580

Even though Ahmed Abbasi’s first paper on marketing was runner up for the best 2020 article published in the most prestigious journal in the field, Abbasi is not a marketing or sales expert.

Instead, Abbasi peers into the future of artificial intelligence (AI) and tries to figure out what all this data can tell us about the human condition. The Joe and Jane Giovanini Professor of IT, Analytics, and Operations in the Mendoza College of Business even named the lab he co-directs after HAL, the seemingly omniscient computer in Stanley Kubrick’s “2001: A Space Odyssey.” Abbasi employs the acronym for his Human-centered Analytics Lab rather than a God-like intelligence.

Funded by the National Science Foundation, HAL has used big data and AI to solve problems ranging from creating early warnings about drug interactions and car safety to improving the natural language processing of call center computers, including how to remove unintentional bias in the algorithmic models.

“Most human behavior now requires a social and technical lens to see and understand it, because we measure behavior through this digital measurement framework, and it’s more complex than it’s ever been,” Abbasi says. “We have access to more information and data than ever before, but at the same time making sense of it is becoming more and more difficult as well.”

Keeping a clear focus on the ethical dimensions and human impact is central to Abbasi’s view of the intersection of people with AI and big data. “To study these problems,” he says, “is really trying to understand the human condition in an increasingly digital world.”

So it was only a matter of time before Abbasi turned his analytical gaze toward the marketing world, where collecting reams of data is the coin of the realm.

“Understanding consumer behavior has become much more challenging because there’s so much more complexity,” he says. “It’s no longer the traditional marketing, more than just a focus group. That’s why in marketing, a lot of machine learning and analytics is now being used.”

His new study, published last year in the Journal of Marketing with three co-authors, found that consumers take very different paths, using different online channels, depending on customer intent and the type of purchase they are making. The title is “Path to purpose? How online customer journeys differ for hedonic versus utilitarian purchases.”

Though born in Chicago, Abassi began his own path in Virginia, where he attended elementary and high school in the suburbs of Washington, D.C. After earning undergraduate and master’s degrees at Virginia Tech, he went to the University of Arizona for his Ph.D. work in artificial intelligence.

He began his faculty life in Wisconsin before returning to his roots with a position at the University of Virginia (UVA). More than a year ago, Abbasi came to Notre Dame because he said his vision for human-centered analytics aligned with Mendoza’s approach to business, which focuses on contributing to human flourishing.

Web Ahmed Abbasi Headshot

“I think we have something to offer in terms of training the next generation of practitioners and scholars, and we’re in some ways uniquely positioned to do that because of our mission at Notre Dame,” he says. “More than ever, students and even leadership want faculty to be thought leaders that look at the humanistic aspects of business as well, so it’s exciting.”

The origins of Abbasi’s foray into marketing began at UVA, where he worked with two of his co-authors. The final co-author helped found comScore, which collects data about online consumers similar to how Nielsen tracks television viewing. From millions of anonymized computer users, comScore has a group of about 100,000 users, statistically sampled to be representative of the U.S. population, who have agreed to have their browsing behavior quantified.

This paper uses all internet activities from 4,356 consumers with 22,751 purchases that account for $1.2 million in sales over two years.

“We’re talking massive amounts of data that we’re analyzing,” says Abbasi. “We’re not only able to analyze 40 retailers and 20 product categories, but also thousands of transactions across millions of browsing sessions.”

Abbasi said his contribution mainly consisted of leveraging the skills of the team – in marketing, search engines, machine-learning, and consumer behavior – and creating a framework to analyze the data.

Today’s consumers don’t usually go online and immediately purchase a product from the first website they visit. Instead, they visit many different sites – often switching from laptop to phone to tablet – over the course of up to a few weeks. This “customer journey” is the series of actions taken to arrive at the moment of purchase.

Product managers, marketers and retailers have understood the distinction between utilitarian and hedonic purchases for a long time. Milk or toilet paper are considered classic utilitarian purchases, while buying a board game or fine wine would be a hedonic purchase.

It’s less a distinction between everyday items and luxury items than it is between what you need and what you want. Browsing a thrift or dollar store can be a hedonic experience for those that enjoy the thrill of finding a bargain.

Stores have found a connection between, for instance, diapers and beer, where the six pack is a reward for getting the needed item. And retail managers act on these differences and behaviors.

“At the grocery store, the utilitarian items are the things that they always place furthest from the checkout because that’s why people are there,” Abbasi says. “The hedonic things are the things they place closest to the checkout because those are the additive items, the impulse-buying stuff.”

So Abbasi’s team conducted a survey to learn customer perceptions of 115 common retailer categories. They learned that the same product category can be perceived differently across retailers. For instance, Amazon’s jewelry and sports products are considered more hedonic than Walmart’s.

“What we were trying to answer was, how do these digital user journeys impact purchase decisions?” Abbasi says.  “The key angles we considered were: what was the consumer intent and how well was it being leveraged. That’s why our paper title is ‘Path to purpose.’”

In case the data wasn’t already complicated enough, the team tracked six different online channels common in the customer journey: search engines, social media, reviews, deals, product pages on target retailer, and on competing retailers.

“Other researchers have individually looked at hedonic versus utilitarian, but they looked at specific channels,” he says. “So you see this smorgasbord of application and the problem is that if you only look at one channel at a time, you don't look at the interaction effects between those channels. In reality, journeys are not defined by just looking at that one snapshot.”

The analysis found that consumers prefer social media and product sites for hedonic purchases, especially early in their journey. For utilitarian purchases, they are more likely to use search engines and reviews early and deal sites later.

“We find that hedonic purchases utilize social media up to 10% more than their utilitarian counterparts throughout the purchase cycle and are less likely to switch retailer brands,” the report said.

“The reason this is so interesting from a theory standpoint is that this purpose angle can illuminate some interesting patterns,” Abbasi says. “But then also practically, if I’m a product manager or an ad agency that’s been asked to drive traffic for one of these retailers, having these insights can help inform my ad space or my marketing strategy in terms of how customers engage in these different journeys.

“Because the key thing we find is that what matters is very different depending on the purpose,” he adds.

The findings also have practical outcomes for retail strategies. To sell hedonic products, retailers should embrace social media, monitor on-site product views and consider social coupons that feature experiential aspects of their product. Managers selling utilitarian products should prioritize search-engine marketing to get on customers’ radar early and invest in automated benchmarking tools to keep their products competitive late. Selling both types of products for Black Friday should begin earlier to align with the journey customers take to get to purchase.

Abbasi said the kinds of analytics performed for marketing are not so different from his other projects. The journey and channels may be different, but similar analytics can still yield valuable insights. For instance, he has analyzed multiple online channels to amass auto user reports that could identify problems like sticky pedals in certain car models.

“What we find is, you can use this information not just to understand the journey and the steps, but rather as a signal,” he says. “Is it an early warning? We’re finding that you can actually detect these events years in advance, compared to the federal databases.”

The similarity in analysis framework may explain how Abbasi’s first paper in the field of marketing became a finalist for the Hunt/Maynard Award, which recognized the 2020 Journal of Marketing paper that made the most significant contribution to marketing theory. He could have stuck with what he knew, but his curiosity opened new doors.

“It’s very easy to succumb to our availability bias and just focus on the data we have,” Abbasi says. “But I think this paper shows we really have to keep the focus on ‘why’ questions.”