Episode Transcript
Will Fraser: Rachel, welcome back to the show. It's great to have you here. Hard to believe it's been almost a year since we spoke and you helped us christen this show as our first guest. Thank you for coming back.
Rachel Gershon: It was such an honor. Thank you so much for having me back.
Will: We've been chatting recently and it sounds like you've got some really interesting work you've been doing. I'd love to learn more about what cool things you're exploring and finding. Before that, for any listeners that haven't heard the first episode with Rachel, I really recommend going back to listen. Some of the content in that episode has made me sound the smartest over the last year. But Rachel, what's been going on for the last year?
Rachel: Along the same lines as the research I told you about last time. That work was on what we called prosocial referral incentives. We were interested in consumer behavior in a social context. People want their friends to think highly of them and so they are even more likely to send a referral if the reward goes to their friend and that friend is even more likely to follow through. So we were interested in this social referral reward, and we've now moved on to some other referral reward research where we're interested in what is the value of a referred customer. There's a lot of other great work finding that customers who are referred are more valuable. We wanted to just add to that research. So we started by replicating it and then extending it. This is some recent unpublished work, which I'd love to share more with you either now or later. This work is with a brilliant friend and colleague, Zhenling Jiang, who is at Wharton.
Will: Nice. It's interesting you mentioned this previous work. In my readings in the space, a lot of the previous work is, I'll call it extremely previous, has been done some time ago. And you kind of just glossed over that kind of reproducibility. Help me understand that. Was this work generally reproducible?
Rachel: Well, I'm so glad you brought that up. You know, it's hard to publish research that's just replicating previous findings. That's getting a little bit better. There's been a real push for just replicating previous findings, making sure that it generalizes to other contexts. And we were excited to be able to first do that in this paper and then also add something novel. We're working with this large mobile technology firm and we were able to replicate that people who'd been referred made more purchases. So we're using that as kind of a proxy for their overall revenue. They're more active, they make more purchases. But you're right, this previous work was primarily in the context of a bank, and it was done quite a while ago. So it was exciting to be able to see that replicated.
Will: Yeah. That's personally really exciting for me because probably 10 years ago, I was reading, I believe, the paper you're speaking about, which is in a German bank setting. It was always interesting, but having that ability to actually verify it culturally, historically, technologically - there's all these questions that I've had over the years. I know you guys extended it, which is really exciting. We were talking before the show about some of those tidbits and new finds. But I think that in itself is important because there's just so many marketers out there that believe it and want to believe it. But the engineer in me wonders if it's true.
Rachel: And there's some interesting context specific findings. It's a bank and therefore some of the things they found about churn, where they found that people, once the person who had referred you left the bank, you also became more likely to leave the bank. There are reasons to believe that might be kind of specific to a bank. Like if you and your partner are using the same bank, you might both leave. Whereas if you have a product that you're using and maybe you referred it, but then you learn that you really like it, then you might be more likely to stay using that product. So we actually didn't find this likelihood of churning in our data. That was one interesting piece - in this context people, once their referrer had left, they weren't any more likely to leave, which is really important.
Will: Yeah, that's really important. It's interesting that you mentioned that social connectedness and how the relation might matter, the product might matter. Broadly speaking, if you're talking data from large scale mobile applications and consumer behavior, once you're introduced to it, the departure of someone else doesn't take you away. I think that just helps speak to this idea we've talked around without necessarily the full backing to support that the lifetime value of a referred customer should be better. There's always been that question about the influence of that referrer.
Rachel: Right. Another nice thing we were able to do here - there's two pieces of the role that the referrer might play. One is social enrichment and the other is social validation. Social validation is that we trust our friends and if they like a product then we might feel more validated in using this product. The social enrichment is the continued aspect. If my friend refers me to this bank and I have questions about how to use the bank and customer service, that's the social enrichment piece. A clearer example of social enrichment might be something like gaming online with your friend - if your friend refers you to play an online game with them, the experience will be enriched by playing it together. And so at least in our context, we found that this social enrichment did not play as much of a role long term as we might have expected.
Will: That's fascinating and I'm just glad there are people out there really trying to test this historic knowledge and help us push that forward. I don't believe humans have changed that much in time, but I believe a lot of things have changed and more clarity can be gained.
Rachel: So just checking everything, right?
Will: Years ago when I started working around customer marketing and referrals, I remember visiting a family member and in their display cabinet they had this old cookbook from say early 1900s. In it was an ad for flour with basically a coupon you could give to a friend. I laughed - there is a referral program, literally probably from around 1910. It's been going around for a while. But having the data we have access to now just makes it such a more interesting area of study to prove how to get it right.
Rachel: Right. They've evolved so much and there are now infinite types of programs. I just experienced a referral reward program I had never seen before in my personal life - an anonymous referral. I received a referral from a friend, but it just said "a friend has referred you to this." As a researcher it was so interesting because I thought from my perspective, I'd like to know who this friend is, I'd like to ask them do you like these shoes? On the other hand, from their perspective, maybe it's just really easy to not worry about reputational concerns of referring. So maybe you're increasing the referral side, maybe decreasing the recipient side. So many new options.
Will: I would love to know the details on that one because almost everyone I speak to wants to go the other way - they want to hyper personalize it to associate that credibility of the referrer. But stranger things have been proven to work in marketing. So would be very interested to see how that plays out from a study perspective.
Rachel: Oh yeah, I will almost definitely be looking into that. Definitely interested in the anonymous component. From the recipient's perspective, that's the whole purpose of the referral reward. I mean, there's the targeting that your friends know you well, maybe this friend knows I would like these shoes. On the other hand, there's no benefit for me of knowing this friend for validation or enrichment.
Will: So I don't want to take all the good bits out of this new work you've been doing, but generally speaking, any other high level insights you want to share with us today or do you want to save that for maybe next time?
Rachel: I could give you a bit of a sneak peek, which is that we found - and I think we might have discussed this very briefly last time - but what we found in this data is that when you just look at a customer's value, when looking at the referred customer's value, you're underestimating the value of that customer if you don't take into consideration their likelihood of referring. What we found is what we're calling contagious referrals. Referred customers are not only more valuable, they're also more likely to refer other people. They're about 30% more likely in our data.
Will: That's sizable. That's a sizable influence.
Rachel: Totally. We found that in this large field data set. We replicated it in a field experiment and in several randomized studies. And we find that it makes a large difference in the value of these customers, which we already knew were valuable customers. So referred customers are also bringing in more valuable customers.
Will: That's really interesting. To me that brings up a question - it sounds like there's an advantage to actually start a referral program early with the idea of establishing that referral contagion. We've called it internally a culture of referrals. Each one's going to accelerate the program as you reach that critical mass. But 30% is a major move.
Rachel: Yeah, I mean, I don't want to give too much away, but another piece we found was that if you just remind people that they joined through a referral, that also boosts their likelihood of referring. So you have these referred customers who are already more likely to refer, and some have been around for five years, and it's maybe not super salient that they joined through a referral. Just reminding them "hey, you joined through a referral, refer your friends" also increased their likelihood of referring. That was really exciting as well.
Will: Yeah, it's interesting because we've talked to several people who discussed this idea about referral likelihood as a time continuum from when you started with the product to where you are now. That's an interesting piece - does reminding them almost reboot them to somewhere earlier in their lifetime as a referred customer that they now remember that and act accordingly? That's really cool findings.
Rachel: And one important piece of this is that we can also control for the amount of activity. Sure, early on customers are more active with the company, but even controlling for that activity, we continue to see this effect.
Will: Awesome. Now, I know I don't want to steal all the thunder from this new paper coming out, but before the show you mentioned some other really interesting work that I think would interest our listeners in the customer marketing space. Maybe you could give us an overview of that work?
Rachel: Sure. I have another project called "Friends with Health Benefits." This project is about what we call tandem incentives. We found people were more likely to go to the gym if we incentivized them to go with a friend. We had people sign up with a friend, and either gave them a reward every time they went to the gym or only when they went with a friend. We found a 25% increase in their likelihood of going to the gym if they had to show up with their friend. The next step we're interested in is do people predict this? Are they aware of this level of accountability that increases their likelihood of going to the gym?
Will: What sparked this question? What made you wonder if this would work or not?
Rachel: Oh my gosh, well for me it's what we call "me-search" where I'm just like, I can't get myself to exercise without a friend. If I tell someone I'm going to be at the pool at 4pm, meet me there, then I will absolutely show up. But I need that accountability. This is one of those cases where I wondered, is this true for other people too? Can we add this hurdle of having to show up with your friend and get people to actually go more often? There's a lot of good work on the role of accountability in motivating behavior. So this was just a way of incorporating accountability into an incentive scheme.
Will: When you gave people an incentive just for going to the gym versus going with a friend - was it the same incentive? It seems like a higher bar to go with a friend than yourself.
Rachel: Yeah, it was the same incentive across the board. This was just $1 - this was done with students. We gave them $1 for going to the gym. We actually ran this study twice. One time we had a $1 incentive and a $3 incentive. We found there was less of a difference - the $1 incentive basically brought it up to the level of the $3 incentive. Having to go with your friend made this incentive a lot stronger, but once you got to a larger incentive, people were about equally likely to go. So this seemed to work a little bit better with small incentives.
Will: Out of curiosity, since we're trying to get people to go with a friend - did you follow any of these people after the incentive structure ended? Were you able to see if it continued, or is that maybe for future questions?
Rachel: Good question. We didn't actually find any long term effect, unfortunately. That's pretty typical of gym incentive studies. We did this for one month and who knows, if we'd done it for three months, maybe we would have found a long term effect. It's not clear this was enough time to establish a habit, but that was what we were hoping for - create this incentive scheme, remove the incentive, but they still have their friends to go to the gym with. One thing to note is we ran it in February 2020 and so - nothing going on, right? We got so lucky, we ran the study for that month. Then the gym closed March 15th. So we had a couple weeks after the incentive where we didn't find an effect. But as you can imagine, there were other things going on that possibly played a role.
Will: What I like about this, and you alluded to this before the show, is I think it has real applications in consumer loyalty programs.
Rachel: Totally.
Will: I can immediately see this coming together when looking at - is it playing a game, multiplayer game, is it travel? There are lots of applications where we can try to get that social commitment to assist us in bringing two people or maybe even more to participate in something. I've spent enough time playing video games to know that if you can incentivize everyone to get together and play at that time, it might be an interesting way to take one key player who's really keen to have a reason to drag their 2-3-4 friends back into it.
Rachel: Totally. You can imagine how we did it as health incentives program, but for any gym, yoga studio, gaming - if you wanted to use this as a kind of referral program, like "bring a friend and if you both come, you both get a free class," you could absolutely imagine how the accountability would still factor in.
Will: Yeah, that's awesome. So what's next? This is multiple great reports and findings. Do you have the next project lined up or the next big question to ask?
Rachel: Oh, we have so many thoughts on what we'll run next. But I really am interested in this anonymized referral. I think that's where my head's at right now. Understanding are there contexts in which this works? Maybe more stigmatized contexts, maybe more uncomfortable contexts where people would be more interested in referring if they could remain anonymous. And then also just continuing to look at different incentives for referrals. From my previous project, we looked at these prosocial referral incentives, but we also found they're about as effective as a shared incentive. We're still really interested in figuring out what it is about the shared incentive that might be working. And you can see how that ties into the gym paper as well.
Will: Yeah, for sure. This is always interesting to me - how much time people look at incentives. Because until you're into it, you don't really think about it. But when you look at a referral program, the majority of your cost is in rewards. When we look at companies, they might think about people time, software costs, but the cost for your rewards could be 20 times the cost of everything else together. So getting them right, not just for their effectiveness but for their cost effectiveness - constantly a question we're getting asked. We can point to what we've seen in our experience, but the research you can do is just at a depth that is wonderful for the market to see and truly understand maybe the why, not just what others have done before.
Rachel: One interesting thing in studying for this most recent referral project - one of the studies we ran, we randomized whether people, in a hypothetical study, either received a reward or not for sending a referral. We asked them to imagine they either joined through a referral or through an advertisement. We were trying to see if we found this boost in referrals regardless of whether there's a reward. But interestingly we found no difference in whether people received a reward in their likelihood of referring. Is this a hypothetical study issue? Is this because we asked them to imagine their closest friend? Maybe people would be willing to refer one person, the first person that comes to mind, but not the second or third without a reward. This was unexpected because we know referral incentives work in the real world.
Will: Right, right. That's really interesting. The more data and research we get on this, the easier it is for us to help those many customer marketers out there who are trying their best. We're all fighting to make sense of it when we do it day in, day out. There's a lot of people looking for the right answer, and I think at this point it's still pretty hard to give that answer.
Rachel: Yeah.
Will: This has been amazing information. Not that you have to, but anything else you've found lately you want to share?
Rachel: Nothing coming to mind right now, but we should keep talking because I'm sure you have a bunch of great ideas about what we should be studying next. It's really interesting to hear that you were aware of this other research and would love to hear what you think.
Will: Once upon a time I spent some time in academia. They kicked me out and said maybe you should go work in the world. To be fair, that also happened in my development career. So maybe not a good thing. But thank you so much. I always love learning these things and I've got a lot of big dumb questions I'm happy to share.
Rachel: Any point as well.
Will: Yeah, well before we break here, I just want to ask - where can people find the research you're doing? Where can they connect with you and follow the great stuff you're doing?
Rachel: Well, they can look at my website. I have links to the unpublished work I've talked about today, which is called "Referral Contagion," as well as the paper I spoke of last time. That is rachelgershon.com. Let's see where else... Yeah, that previous paper was published in the Journal of Marketing Research and it's something about prosocial referrals.
Will: Perfect. Well, first of all, I'm jealous that you managed to get your name as a domain - mine's a rugby player. But thank you very much for all your time and insights here today. For our listeners, we will link to those sites Rachel just mentioned. Thank you very much, Rachel. Cheers.
