How Consumer Attention Impacts Decision-Making

April 27, 2021

The marketplace provides abundant choices on any product we may be looking for – but then overwhelms us with information about them. That is why attention has become a scarce resource for consumers. In a Marketing Science Institute expert curation, Tulin Erdem, professor of business and professor of marketing at New York University’s Stern School of Business, examines the impact of consumer inattention on search and choice, finding that inattention might largely drive people’s search and decision-making.

Erdem recently joined Marketing Science Institute Executive Director Barbara Kahn, a marketing professor at The Wharton School at the University of Pennsylvania, to discuss her findings. Erdem discussed the papers, “Prior Information and Consumer Search: Evidence from Eye-Tracking,” co-authored with Raluca Ursu of the University of Chicago and Qianyun Zhang of New York University, which was produced with the support of MSI; and “Attention Trajectories Predict Brand Choice,” co-authored with Ana Martinovici of Erasmus University Rotterdam in the Netherlands and F.G.M. “Rik” Pieters of Tilburg University in the Netherlands.

Listen to the interview:

An edited transcript of the conversation follows.

Barbara Kahn: Before we get into the specifics of this research, why don’t you describe the kinds of things that you’ve looked at, and position this type of research in that general research stream.

Tulin Erdem: I have been working in the general area of choice modeling and consumer decision making for a very long time. I have different research streams, but one important one has always been how individuals make decisions, and especially choices in the marketplace.

I utilize many different types of data, from transactional data like scanner and panel data, to survey data, to conjoint analysis studies to uncover consumer preferences per sections, learning about the attributes in products, and ultimately their choice behavior. I look at their customer journeys.

I come from that kind of perspective in terms of choice modeling. About six or seven years ago I started also focusing on the reality that consumers’ attention is getting more scarce than ever. There are several reasons. One is that we have many more options in the marketplace. Secondly, now product categories are getting much more complex. For example, even in a category like smartphones, there are thousands of features that a consumer could consider.

Our typical models were really not designed to get at those issues, because you need about seven or eight attributes. If there are thousands of them, how will you model these things? Most of these models assume consumers pay attention to all the information that is out there. Given the complexity of the decision-making environment and the information overload, consumers don’t have the time, the willingness, or the ability to collect all that information. They need to make a judgment in terms of what they attend to and what they don’t attend to.

Kahn: A lot of your past work and a lot of research on decision-making assumes that people are taking in this very complex decision-making process, and weighing or trading off the important weights of one attribute versus another, and going forward, making more rational decisions. But in today’s world with enormous information overload, when everyone has six screens open on their computer monitors, that is not the right model for decision-making anymore. Now it’s a better model to try to predict what people are paying attention to and what they are not paying attention to.

Erdem: Precisely.

Kahn: I know you’re going to tell us you use eye tracking to find out literally what people are paying attention to or not, but when you went into this work, did you have a hypothesis as to what you thought would drive attention? Or did you let the data tell you what it was?

Erdem: It was a little bit of both. I have two papers in this area. In one of them I had a formal hypothesis. There is, of course, the literature utilizing eye-tracking data, so I was building upon and extending the literature. The basic tenets are that it is rational for the consumer to make these expected benefit-expected cost calculations, so to speak, and choose not to pay any attention to certain things.

That is to be expected. But to be more precise, to make more nuanced hypotheses, we did some of that before looking at the data, and we discovered other things after having looked at the data.

Kahn: Tell us what you discovered.

Erdem: One thing that was interesting with the models we built is that we can predict choices quite early in the process. There was some literature that said toward the end of the process it was possible to predict what the consumer would do. But with our models, my co-authors and I were able to predict choices pretty early. For example, if the task is X minutes, or if consumers take X time to decide — whether it’s 15 or 20 minutes to go through some information — if you divide that into four quarters, we were already predicting quite well in the second quarter what they would do.

Kahn: Wow.

Erdem: That is very revealing, because it already tells you how consumers come with certain preferences already in their decision environment, and what they attend to is a function of their preferences. But at the same time, there is a reverse causality or feedback loop of what captures their attention, and then drives their preferences. The moment you are precisely measuring their attention to eye tracking, which is the best way of measuring attention, you can predict what they will do very quickly.

Kahn: Are you saying their preferences form what they pay attention to, and that’s going to drive what they ultimately choose? But it’s also true that what gets their attention seems to also form their preferences going forward?

Erdem: Exactly. So there is this feedback loop, this loop relationship between attention and preferences. Both of them together affect decisions. Another thing we found, we already expected that prior experience, prior brand ownership would affect what consumers do. But interestingly, what we found is that consumers indeed pay attention more to the brands they own, but once you capture that effect, there are no other remaining effects of past brand ownership. That was one of the interesting findings in that paper.

Kahn: What was the focus of the other paper [about attention trajectories and brand choice]?

Erdem: The MSI paper utilizes similar data but focuses on a slightly different issue: How do people choose what to attend to? We have consumers’ eye fixations, and we are modeling it as a search model. I can decide to look at, let’s say, if I am looking at smartphones one consumer may focus more on Apple, another may focus more on Samsung. But also people may focus on a certain attribute more. For example, prize, or screen size. So that model looks at how consumers prioritize information before they even start deciding or collecting information, and how that affects, along with other things, that search process, that consumer decision of what to attend to and what not to attend to. Once you model this, you can correctly explain consumer heterogeneity in how they behave, because different consumers behave differently.

Some consumers may just stick to their own brands. Others may be more variety-seeking, so to speak. Some consumers may focus on one attribute, others may focus on multiple attributes. Once you look at this whole heterogeneity of the process, you can see what matters to different consumers. And then once you estimate the model, you can create simulations to see how you can affect consumer behavior.

For example, if you are a brand, and you are seeing that people really don’t pay much attention to you, what you should be doing? Maybe we find that a certain attribute attracts most attention from most consumers. Maybe you have to focus on that attribute.

Or if you know you are better on a certain attribute, you can do certain things on your website [to highlight that.] For example, in that paper we discuss recommendation systems, and the implications of this attention for marketers and managers, how to use that information to design their web flow.

Kahn: What you’re describing makes me think of the behavioral change literature, where people talk about increasing importance weight. In other words, if you were good on a particular attribute, you could increase the weight of the importance of that attribute so that you would win.

You’re suggesting a parallel process, not based on importance weights, but based more on attention. That is somewhat provocative because the other way is very much about rational decision making. Here you are almost talking about salience, or driving attention, and not necessarily attributes that would literally change persuasion mechanisms.

Erdem: Exactly. It’s a bit of both. For example, again, the MSI paper, it is a learning model. We are also looking at how people learn. That’s what the old literature couldn’t do, because there was no data. To even learn, you need to pay attention. Part of that attention, as I said, is of course preferences. But some of the attention also happens for other reasons. So even learning needs to start with attention.

Kahn: So I was making it too simple, and you’re going to scale me back a little. Yes, some of that old stuff is still relevant, but we didn’t put enough weight on attention. That’s a very interesting finding, I think. People will find that useful going forward. How important do you think eye tracking is as a mechanism? How much weight should marketers put on using eye tracking?

Erdem: I think it’s a great data source. All different data sources, of course, are complementary; they are not substitutes. But eye tracking, especially in terms of understanding consumer choice processes, and of course attention, is a great resource. Both in marketing, consumer decision making, consumer behavior, but also in behavioral economics.

Lots of people called for use of eye tracking data to test some of the theories developed in behavioral rational attention theory. Because indeed, in this kind of environment, it’s quite rational. It’s rational for consumers not to pay attention to everything. And what better way to understand what people attend to, what types of things grab consumer’s attention more, keeping even preferences constant [than eye tracking].

We just say that of course people will look at more things for which they have higher preferences. But even keeping that constant, what grabs attention in these hyper complex decision making environments? Eye tracking … is a very rich data source to uncover all those processes and test what is best to increase the salience and attention.

Kahn: Has most of your eye tracking research so far been conducted online or on a computer? Because now Amazon’s operating these physical stores where there are cameras and sensors in the store, and presumably you can use that kind of behavior … to try to measure attention not just online, but also in stores and in physical environments. Have you started to look at that? If not, can you comment on whether that’s feasible?

Erdem: I personally haven’t started working on that sort of data, but I know colleagues who are working on it. One of my colleagues has been using such data to look at similar issues, even age-old questions of how you decide about shelf space, and how retailers arrange or make decisions about what to put where.

These technologies that are being developed now in a very unobtrusive way will be very helpful for various types of measures for retail business or manufacturers to understand and uncover consumer attention. Not only attention, but also to learn what decision rules people use. Again, in the typical modeling it is very difficult for us to see or measure what is the exact decision mechanism consumers are employing.

Like are they using a view of, I will only look at price first, eliminate certain things, and then just compare the rest? We try to infer, but we don’t have a direct mechanism of seeing that. With attention data like eye tracking, you know exactly where they are looking, whether they are looking at a specific attribute or if they were never looking at price first, for example. Or just the opposite, they only look at prices, and then they start comparing different brands on different attributes.

The heuristics, like shortcuts in decision rules they use, are a great data source to uncover that. And why is that important to uncover? Aren’t heuristics just an academic interest? No, because if you know what the consumer heuristics are, what the decision rules are that they are using, first you can increase their welfare by making the environment easier for them. Two, it also gives you the mechanism to also interfere to then to give the right messages at the right time of that decision process.

Kahn: I think that this is more important than ever. As we said in the introduction, because now you’re being bombarded by information all over the place. As we moving to multiple screens in our lives, getting the consumer’s attention is probably one of the most important things a marketer can do.

By using MSI.org you agree to our use of cookies as identifiers and for other features of the site as described in our Privacy Policy.