Wharton's David Bell discusses his new research on how 'social learning' influences online sales.

Word-of-mouth testimonials from consumers are effective in driving online sales. But these signals are even more powerful in communities where people have closer ties and trust each other, according to new research by Wharton marketing professor David Bell and Jae Lee. That means online retailers get more bang for their marketing buck if they target close communities. Their paper, “Social Learning and Trial on the Internet,” was sponsored by the Mack Institute for Innovation Management at Wharton.

An edited transcript of the conversation follows.

Knowledge at Wharton: Your research paper looked at social learning and social capital, and their impact on internet retail sales, and you found some interesting correlations. First, can you define social learning and social capital for us, and then tell us about your research?

David Bell: I’d be happy to. First of all, let’s start with social learning. Social learning is just the idea that individuals communicate, either directly with each other, or they learn from each other via observation. So, you might have said to me this morning, “Hey David, you’ve really got to go over to Starbucks. They have the best espresso.” And I might take that information, learning it from you, and go over there. Conversely, I could just see you holding a cup of coffee with the [Starbucks logo of a] mermaid on the outside, and that might make me go get a coffee myself. So, it can either happen — social learning — through conversation, or through direct observation.

… Social capital was a term that was coined some time ago in the social sciences literature, probably popularized by [Harvard professor] Robert Putnam. He actually worked on President Obama’s first election campaign, I believe. He wrote a book called Bowling Alone, about social capital in America. The essential thesis was that people — at least, when he looked at them in the United States — [over time] tended to go to church less, join the tennis club less, be less involved in community activities. This is [exemplified by] the metaphor, I suppose, of bowling alone.

And, to really bolster this theory, he went out and collected a wide variety of data from about 30,000 households. He would ask you questions like, on a scale of [say] one to five, how much do you like your neighbors? You know, from ‘I really detest them,’ to … ‘I really enjoy them.’ And also, [he looked at] the frequency of interaction. So, social capital is really about the notion of interaction and trust; how much of those two things co-exist in a community?

Knowledge at Wharton: How do social learning and social capital play a role in terms of online retail sales, according to your research paper?

“The idea that people might physically observe each other or have a conversation about the product — that was responsible for up to half of the sales.”

Bell: It’s a little bit of a step from one to the other. We looked at a product category, fashion apparel, which is a socially observable product, first of all. So, if my friend is wearing a particular pair of glasses, or a jacket, or pants, I might comment on it and he might tell me where he bought them. The idea is that the product has to be socially visible, and potentially something about which a conversation could take place.

Where social capital comes in is, we wondered whether or not the effect of information that was shared in a community differed, depending on whether it was a community where people didn’t really know and trust each other too much, or it was a community where they did know and trust each other.

Knowledge at Wharton: What kind of data did you look at, and how expansive was the sample?

Bell: We took data from a men’s fashion retailer called Bonobos. It was started by two MBA students back in October 2007, essentially to sell men’s apparel online. They now have physical stores in about 30 cities, and they also have a relationship with Nordstrom. So, they’re both online and offline. For our listeners, the jargon is they’re a digitally native vertical brand, which means that they were born on the internet, but have offline, too.

They provided us with their national sales data. So, we know, for example — and it is obviously anonymized — that a certain customer in a certain location purchased particular items at this time and for this price. And so, we had basically a footprint of the entire United States, in terms of sales, coming from the company. And then, we were able to match that with the Social Capital Community Benchmark Survey — the official name of the data that Robert Putnam and his team collected — and that covers about 30,000 households, if I remember correctly.

Knowledge at Wharton: What was the purpose of your research? What were the hypotheses you were trying to test?

Bell: The main question of interest in these studies often is — since it’s an internet retailer, so by definition it has a very large footprint and covers the entire United States — how might their sales evolve over time, and over location? If different locations have different characteristics — different kinds of people, different kinds of offline stores — does that have any impact on the sales? A number of authors — not only myself, but at other institutions around the country — have addressed that question.

In this particular instance, our key variable of interest was the cohesiveness of the offline community. Did that have any impact? We found two things there. First and foremost, we found that for this particular company, social learning was important in generating about half of all of their sales. So, just the idea that people might physically observe each other or have a conversation about the product — that was responsible for up to half of the sales. It was particularly important for sales that occurred later in the process.

The idea behind that relates to old ideas, and diffusion of innovation. So, if [my friend], for example, is a real innovator, he might just go out and buy Bonobos. He doesn’t need any social validation, or anyone to tell him about it. He’s maybe the kind of guy that queues up in the morning for the new iPhone. Someone who comes later in the process may come because of social influence, or social cues. So, that was the first finding — this idea of offline information transmission was driving online sales.

The second piece, the social capital piece, was really fascinating. In communities where people knew and trusted each other more, the diffusion happened more quickly. Now, it was not the case that if you live in a neighborhood where you trust your neighbors, you’re more likely to buy online. It’s not that at all. It’s more subtle. It says that if you live in a neighborhood where you know and trust people, when information is shared in that neighborhood, it has more veracity. In terms of economic jargon, the signal is more powerful. So, if two people in your neighborhood mention Bonobos, that’s enough for you to go and act on it. In my neighborhood, where nobody likes each other, nobody trusts each other — things that are communicated offline have less veracity, and to get the same impact, I might require five signals. So, that was the key finding there.

Knowledge at Wharton: What are the practical implications of your findings, and how can internet retailers benefit from them?

“If you live in a neighborhood where you know and trust people, when information is shared in that neighborhood, it has more veracity.”

Bell: There are really two practical implications. One is just to reinforce the idea that in internet retail, the physical location of customers really makes a big difference to sales, even though an internet retailer may be flat — by that I mean in terms of access. Anyone in the country can go to a website, buy a pair of pants, pay the same price, get the same shipping, and so on. But, their propensity to do so varies dramatically by physical location, because different communities have different offline options. They have different kinds of people living there, and the people living there have different preferences.

So, those three things explain a lot of the geographic variation in sales. Because that’s such a fundamental aspect of online retailing, we’re always encouraging internet retail companies to think about what are the geographic factors that matter. We just so happened to have keyed onto one, which is the cohesiveness of the offline community. And so, you’re Bonobos and you’ve got some marketing budget to spend. You would be better off dropping that money into communities where you know there’s an offline cohesion, than communities where there’s not.

Now, that sounds a little bit esoteric. Gee, how do we know that? Well, the managers could get hold of the Social Capital Community Benchmark Survey, and fill out the forms, go through university channels. That would be arduous. Also, those data do not cover the entire country. They only cover a sample. So, what Jae and I had to do to make it managerially actionable, is to try and think of a proxy for the extent of offline trust and interaction among the target group, which is fashion-forward, 25- to 45-year-old guys. We had a little bit of fun.

… What we did was, we thought, gee, what would be a proxy that would be easily scraped from the internet for this idea of trust and interaction among males in that demographic? What we found was that if we collected data very easily on the number of bars and liquor stores per capita in a particular location, that actually was a proxy for young guys, fashion-forward trust and interaction. We tested that out statistically, and we talk about that in the paper.

Knowledge at Wharton: Are you saying that urban centers are better for online retailers?

Bell: Well, it’s very interesting. There’s a fellow called Chris Anderson, [the editor-in-chief of Wired magazine] who popularized an idea [in a book] back in 2006 called The Long Tail. [He espouses] a fascinating and fundamental idea for anybody who cares about the internet. He says that, in the world of the internet, where you can essentially offer many more books, many more videos — there’s no constraint on variety if I’m a seller. Of course, sales from things that are really popular, like Amazon sales of Harry Potter books, are important.

But there’s a bunch of stuff that they sell in ones and twos that actually adds up to a lot. And those ones and twos you would never physically hold in a store, because they just don’t justify the shelf space. It turns out … there’s an idea of a spatial long tail. What that says is if I’m an internet retailer, yes, I have to hit the big locations first. And I’ve found — if you look at Diapers.com, WarbyParker.com, Bonobos.com, etc. — the large urban centers always predominate initially, in terms of sales: New York, Boston, San Francisco, Philadelphia, Los Angeles, Chicago. But, in order for those businesses to really survive, and in fact, to thrive, they must also pick up customers in what are some of the more remote places, sort of the ‘tail’ locations.

Those ‘tail’ locations tend to be geographically far apart, but similar in some other socio-demographic dimensions. So, there’s a community somewhere, located in Texas, that actually is not that different from one that’s somewhere in Nebraska, and identifying them becomes very important.

Knowledge at Wharton: Is it easy for businesses to find out which communities those are?

“You would be better off dropping that [marketing] money into communities where you know there’s an offline cohesion, than communities where there’s not.”

Bell: The good news is there’s just so much publicly available data out there now. There are things that we can get from the [U.S.] Census. Also — I’m not giving a plug, necessarily, for a particular organization — there’s a company called Esri, and they collect a lot of geographic data. They repurpose data from the Census, and so on. So, there are those kinds of suppliers, who are relatively low cost. You can come up with 50 to 100 variables that would describe one community vis-a-vis another.

Knowledge at Wharton:  It sounds like word-of-mouth and social capital are really important things in a community to drive online retail sales. What can internet companies do to drive more of that positive word-of-mouth?

Bell: One thing that they can do — and this is really outside the purview of this particular research, but relates to ongoing research that I’m doing — is that they can think hard about opening physical locations themselves.… The CEO of the very company that we studied in this paper, Andy Dunn, back in 2009 to 2010, [has said], “Gee, in this internet age, why would anyone need a store? Isn’t that a waste of space, and time? And, you’ve got a whole inventory [you can display online].”

Ironically, what’s driving his business right now are physical stores. But, they’re different [from traditional brick-and-mortar stores]. They’re guide shops. They’re very, very small-footprint stores where customers have great experiences getting to know the brand and the product, [as well as their own tastes,] but there’s nothing they can take away. [Customers place orders at the store and the product is shipped.] This kind of footprint store is being copied and emulated by a number of online retailers.

The reason it’s so successful is twofold. One, it gives people the chance to have a deeper experience. Second, the footprint of these stores, in a digital age, is much bigger than it was in a pre-digital age. To give an example, if a friend and I have a store back in 1990, and 100 people visit that store during the course of the day, then, by the end of the day, perhaps 100 or 110 learned about what our store was about, because 10 people told their husbands or wives. In 2017, perhaps 100,000 people learned about that store, because you’ve got some fellow that walked in who has 50,000 followers on Instagram, and he takes a photo, and he talks about it. The fact that these physical spaces can be amplified digitally also makes them much more important.

Knowledge at Wharton: How is your research adding to the body of work that already exists on social learning, social capital and retail sales?

Bell: I think the real twist here is we’re showing that an offline interaction — something that happens physically between people — matters a lot for the internet. Early thinking around the internet was that the internet and the offline world are substitutable. Gee, I’d better not let Black & Decker sell their drills at Blackanddecker.com if I’m Home Depot, because that’s going to hurt me; these are substitutable things. But, the emerging view is that offline and online are complements. It just becomes a matter of [figuring out] what things you do offline, and what things you do online. There’s absolutely no doubt that what goes on in the offline world influences sales and other activities in the online world.

“There’s absolutely no doubt that what goes on in the offline world influences sales and other activities in the online world.”

Knowledge at Wharton: How will you follow-up your research?

Bell: Academic research often proceeds incrementally and fairly slowly. One thing that I’m currently doing is looking at the impact of these physical, offline spaces on online sales. What I’ve come up with so far, with two of my colleagues — one at Kellogg, and one at Tuck — is that these spaces act as a turbocharger, in the following sense.

If we look at customers who were born online — they made their first purchase at Bonobos online, and they have never, ever visited one of its stores — they have a certain trajectory in terms of average sales, frequency of visiting, the number of product categories of apparel that they buy, and their return rate. Then, if we look at customers who, at some point, went in to a physical Bonobos store, we see that those customers, on average, buy more, they return products at a lower rate, they show up with increasing velocity and they buy more breadth of assortment.

This is the notion of customer turbocharging. If you and I had never met before, and we exchange an email, that’s an electronic interaction. It’s a very low-energy thing, right? If I call you on the phone, and we heard each other’s voice, then the energy of our relationship would increase. Now, we’re meeting each other face-to-face, and we’re having a discussion. Our intimacy has increased a great deal.

The next time I receive an e-mail from you, the energy never drops way back to where it started, with the electronic. That’s a nice metaphor that now our relationship has been somewhat turbocharged. So even if we continue our interaction online, we’ve got this great offline experience that anchors it.