Using behavioral science to improve your product | Kristen Berman (Irrational Labs)

[00:00] economic says, [00:01] "Look, people are rational. We make decisions with no emotion. We use lots of computational energy, weigh the pros and cons." [00:09] Obviously, [00:10] That's just not true. It ignores the whole field of psychology. And so [00:13] In behavioral economics, you combine the field of psychology and economics and say, "Look, people make decisions with lots of emotion. We [00:22] our present bias we overweight our our present selves we follow social norms but the good news is that we do these things in predictable ways [00:30] And once you understand how and why people behave, you can start to change it. [00:35] And so behavioral science and behavioral design basically uses those insights on psychology to actually apply it within real world problems. [00:45] Welcome to Lenny's podcast. I'm Lenny, and my goal here is to help you get better at the craft of building and growing products. I interview world class product leaders and growth experts to learn from their hard-won experiences building and scaling today's most successful companies. [01:00] Today, [01:01] My guest is Kristen Berman. Kristen is the CEO and co-founder of Irrational Labs, where she and her team use behavioral science to help companies like Google, Airbnb, PayPal, Microsoft, Fidelity, and TikTok build better and more successful products. In our conversation, we cover a ton of real-life examples of product changes that her and her team helped craft that led to significant impact at the companies.

[01:31] your product, her biggest surprises and insights about human psychology and how it relates to product design, a couple of real-life case studies, including her work with TikTok and the change they made in the product as a result, and so much more. I found this conversation super fascinating and super tactical, and I can't wait for you to hear it. And so with that, I bring you Kristen Berman. [01:54] Hey, Ashley, head of marketing at Flatfile. How many B2B SaaS companies would you estimate need [02:01] their customers? At least 40%. And how many of them screw that up and what happens when they do? Well, based on our data, about a third of people will consider switching to another company after just one bad experience during onboarding. So if your CSV importer doesn't work right, which is super common considering customer files are chock full of unexpected data and formatting, they'll leave. I am 0% surprised to hear that. I've consistently seen that improving onboarding [02:31] one of the highest leverage opportunities for both sign-up conversion and increasing long-term retention. Getting people to your aha moment more quickly and reliably is so incredibly important. Totally. It's incredible to see how our customers like Square, Spotify, and Zora are able to grow their businesses on top of Flatfile. It's because flawless data onboarding acts like a catalyst to get them and their customers where they need to go faster. If you'd like to learn more or get

[03:01] at flatfile.com slash lenny. This episode is brought to you by Whimsical. When I ask product managers and designers on Twitter what software they use most, Whimsical is always one of the most mentioned products, and the users are fanatical. Whimsical is built for collaborative thinking, combining visual, text, and data canvases into one fluid medium. Distributed teams use [03:31] and even feature specs. And that includes thousands of built-in icons and a rich library of templates. See why product teams at leading companies call Whimsical a game changer. Visit whimsical.com slash Lenny to have my own templates added to your account when you sign up. That's whimsical.com slash Lenny. [03:54] Kristen, welcome to the podcast. Thanks, Lenny, great to be here. It's great to have you. You're on something called Irrational Labs. And so to set a little context for folks, [04:04] Can you just talk about what it is you all do at Irrational Labs? [04:07] Yeah. So Arachia Labs, we are a behavioral science consulting and design shop. So we [04:15] Think about all things behavior change. [04:18] So we work on understanding psychologies that drive users, [04:21] and then designing and testing ways to change behavior and drive engagement. And I started it in 2013 with,

[04:31] Dan Ariely, who's a famed behavioral economist. We started with just working with a handful of companies and now we have a team of 20 behavioral scientists working on [04:41] behavior change across industries. Can you talk about some of the companies that you've worked with just to give some examples and then what kind of stuff have you actually helped them with? We started a lot of our work at Google. We were actually embedded within Google for around three years. I even had a bus pass and we started their behavioral economics team. We worked with over 25 teams from like self-driving cars, [05:03] to YouTube. And so that was a big piece of the early work. And then recently, we've done work with [05:09] TikTok or we've [05:10] tried to, it actually accomplished decreasing misinformation share by 24%. We worked with One Medical on how to increase [05:20] doctor setting doctor's appointments right away during the onboarding and that worked and drove it up by 20 percent. [05:27] We worked with Credit Karma on basically helping them increase setup for reoccurring deposits. We did that and increased it by 18%. [05:35] So basically a variety of ways to drive behavior change. [05:40] done stuff with Intuit, Microsoft, LinkedIn, and obviously a variety of startups as well. Wow. Okay. We're definitely going to talk about some of those examples. I'm super curious to hear what you did with TikTok, big company and story these days. Before we get to that, what exactly is behavioral design and can you just define this space for us? Let's start with what is behavioral economics. So basically behavioral economics

[06:03] is a reaction to economics. So economics says, look, people are rational. We make decisions with no emotion. We use lots of computational energy, weigh the pros and cons. [06:15] I mean, obviously, that's just not true. It ignores the whole field of psychology. And so, [06:19] In behavioral economics, you combine the field of psychology and economics and say, look, people make decisions with lots of emotion. We are present bias. We overweight our present selves. We follow social norms. But the good news is that we do these things in predictable ways. [06:36] And once you understand how and why people behave, you can start to change it. [06:41] And so, behavioral science and behavioral design basically uses those insights on psychology to actually apply it within real-world problems. What are some of the biggest surprises or maybe unintuitive results that you've come across working with companies using these insights in behavioral design, behavioral economics? [07:02] Yeah, great question. [07:04] One story that recently happened is actually a surprise that happened to [07:10] The client, the partner was shocked and actually didn't surprise us. [07:14] And so this was with a popular FinTech app. [07:17] and their most requested feature was budgeting. [07:20] So it came through all the support forums, interviews, [07:24] Felt like a table stakes feature. People wanted budgeting. And so if you're the PM there, you're probably like, [07:29] You may talk to a few other users, you may get some confirmation, but you're likely building budgeting.

[07:34] And that's really where they were when we came and started working with them. We couldn't [07:39] move them in another direction, but we did convince them to do an experiment. [07:44] So we convinced them to test and there was three conditions. One was the control, which is just like telling people how much they spend. [07:51] And the other two were variations on budgeting. So two different ways to do budgeting. So big experiment. This is with 10,000 people. So we can really trust the results. And what happens? [08:02] Nothing. So they were shocked. The team was shocked. Remember, this is like a big investment of time and effort. [08:08] Took a lot of energy to build. Zero change in average spend. Zero change in spend variability. [08:14] And so the question is, like, why weren't we surprised? Why were they surprised? [08:18] And so the thinking here is if you map out [08:21] every single step that it takes. [08:23] to do this behavior and the behavior would be reducing your spend [08:26] it's pretty clear that budgeting just isn't going to pull the weight. [08:30] and this is what we call a behavioral diagnosis where you map out every single step so [08:34] You'd have to like know what your budget is. You'd have to know where in the month you are. [08:38] You'd have to [08:40] make a plan to do it, [08:42] You then have to do it multiple times during the month. Like incredibly difficult to do this one behavior. And once you do that behavioral diagnosis and you map out, you try to understand what people actually do. [08:53] versus what they say they will do. [08:55] And by doing that, it becomes wildly clear that we should all be skeptical that budgeting would actually work to change the behavior that you're trying to get to. Yeah, I'm also not surprised that didn't have a big impact. I have so many apps with budgeting features and I've tried many times to use them and it never sticks. It's like, yeah, I'm just going to. It's too much work. It's way too much work. And so some of any kind of work that we put on the user, we should be skeptical.

[09:21] We have to really prove [09:22] that it's worth their time and then measure if they actually do it. Is there a better approach to that specific problem that you found? Or is that too broad of a question? Kind of the first line of defense in any problem solving would be, how would you make it easier for somebody to do? [09:38] And so in the personal finance world, [09:41] really the answer to most things is just default. So it worked for automatic enrollment for a 1K. The reason we have retirement savings in America is because people are defaulted into saving from their paycheck. [09:52] So that would be first line of defense. [09:53] obviously difficult for product teams to execute on, takes a lot of back-end infrastructure. So if we're just solving the personal financial management, [10:01] stuff like a budgeting alternative. [10:04] we'd actually go to making it [10:06] not just logistically easier but cognitively easier so we call them rules of thumb [10:10] Like instead of having to decide in the moment, do I do this thing? And you're weighing the pros and cons, you know, it's kind of typical econ. Like, how much is this $3 worth to me? [10:21] you'd actually make a rule of thumb that says, do I do it or not? So instead of deciding, do you take a lift? [10:26] you know when you're coming home from work you have a rule of thumb that says [10:30] I don't take Lyft on the weekdays. I take it on the weekends. [10:33] Very simple now to actually make this kind of decision. [10:37] It's a heuristic. It's not going to be perfect, but it's going to help you reduce your spend in an easier way, at least in a way that you'd stay adherent to. That reminds me a bit of James Clear's is kind of idea of creating this identity for yourself, where he's like, if you're like, I am a person that doesn't take lift on the weekdays, that becomes kind of this identity component. And that makes it easier for you not to do that. Is that related? Yeah, basically, you know, sometimes we think behavior change is driven by our attitudes, our preferences.

[11:07] I really want to be a runner or like I really like this thing. But in reality, our behavior is driven by what we do. So in order to change behavior, you have to do something different. It's not enough to have a goal, objective. [11:22] to do it and i think that's kind of maybe that's like the big [11:25] annoying thing about the self-help industry. It kind of just tells you it's enough to have a goal. It's not enough. [11:31] You actually need to redesign your environment to change your behavior. Is there any other examples of big surprises or intuitive results from work you've done? [11:39] One big one is actually... [11:41] And I'm sure people on the listening have worked on signup flows before. This is kind of a common... [11:47] question and challenge in product world. And the common thinking is, [11:52] and most often correct, that if you reduce friction to the sign-up flow, your conversion will increase. This is kind of the law that we all live by. And yet, there is one example that, [12:02] One time. [12:03] where this is just not true, that you can actually increase friction and increase conversion. So actually, Lenny, have I ever told you about [12:11] My mother? Nope. [12:13] Okay, so now you're thinking about my mother. Before you were not thinking about, there's nothing about my mother, but I just made you think about my mother by asking you a question. Son of a gun. So when you ask a question, you can insert an idea into someone's head. You can get them thinking about something different. [12:30] And so in a sign-up flow, [12:32] What would you want people to be thinking about? [12:34] you probably want them to be thinking about the benefits that you offer.

[12:38] So if you ask them a question, they can think about the benefits. [12:41] It's going to increase their motivation to get over the hump of the future friction you're going to put in front of them because now they understand the benefits. [12:49] So as an example, if you go to apartment lists, they ask you questions during the sign-up flow that's like, [12:54] what kind of apartment do you want? Do you want a studio or a one bedroom, a five bedroom? Do you want a patio, a basement, a view? Now all of a sudden I'm thinking, I'm engaging with the benefit [13:04] I understand how much supply that they have. [13:07] So if you like you stitch fix or actually trunk club has come out and said they, [13:12] did this and it increased conversion by 133%. So asking these questions can really drive conversion. [13:20] We actually did something where we put a quiz on a partner's website [13:24] And [13:25] It's title care, which is basically a device that lets you remotely connect to a physician, [13:31] So a little bit hard to use, but we wanted to help people [13:34] understand [13:35] the device, and we ask them questions about their medical behavior, their frequency use of technology. And for people who completed the quiz, 53% of them went on to purchase versus 37% who didn't. [13:48] didn't complete. So these type of questions can have a real impact. And so very tactically, the idea here is you add a step to your flow that asks the user some question that isn't actually something you need an answer to, but [14:01] may get them to be more excited by finishing, is that right? Exactly, yeah. And I would say probably most teams can do this. You can replace

[14:09] The carousel, like the carousel, you know, it's something where it tells you the benefits. [14:14] But we want people to actively engage with the benefits so that they're thinking about how you can improve and affect their lives. So you said it's worked a couple times. It probably isn't something that usually works as I experienced. Like sometimes it may work, it may be worth trying, but usually it probably won't work. Is that generally right? I think, you know, we once removed a field that was like an open text field. We're asking people an open-ended question and we improved conversion by 40% page over page. So for sure, [14:43] some questions are harder and you should not ask them. When questions are easier, [14:47] and it's a drop-down multiple choice, this is when you should be trying it. But definitely the recommendation is not to ask a hard question where people [14:57] have to think about it. By the way, for the question that we did [14:59] It was open text, [15:01] the folks on the team thought that it was an easy question. [15:05] But most questions when you're asking people kind of like an open text field is just not an easy question. Right. I hate those open text fields. And then just one last because I imagine people are listening to this like, oh, maybe you should run this experiment. Is the general best practice to ask a question that gets you to think about something that motivates you by finishing? What's like the rough mental model there? [15:25] So I think it's each each. [15:28] product obviously will be different. One example would be like, so if I'm a bill pay app, [15:33] You could tell me that I could pay my bill once a week, every day, every month. I could time it.

[15:41] But a better approach would be to ask me, [15:43] Do you want us to pay your bills immediately or every week or every month? And so what you're doing there is engaging people [15:50] In the setup, [15:51] and actually implicitly telling them we can pay your bills immediately. [15:55] So I would think about kind of combining those two is showing people the benefit, like [16:00] how fast you want us to pay your bills which gets you to think about how fast they are and the settings [16:06] which is actually like you're really doing something to help the app work. Awesome. Before we dive even deeper into more examples, I wanted to zoom out for a moment. And I'm curious, just how did you even get into behavioral economics? You said you worked with Dan Ariely. How did all this come about? [16:22] Yeah, great question. So I met Dan around 2008. And if you don't know Danny Riley, definitely worth a Google. [16:28] And at that time, he had just written his first book, Predictably Rational. And I was a product manager at Intuit, working on QuickBooks Online. So first use, customer discovery. And [16:40] Really after reading the book, hearing the talk, it's just like this light bulb went off and it was like, [16:44] I'm like talking to five to ten customers to inform a feature. [16:49] a product, [16:51] and try to come up with some like insight on human behavior and yet there's this [16:54] whole field that studies this, right? That has papers and thinking and studies why we behave the way we do. [17:02] And so this was like, wow, I can't believe we're just missing out on it, the product community. And so I started working with Dan and we started and now 14 years later, this is this is

[17:14] That's what I do is bring product insights, bring behavioral insights to product teams. You said you were running an embedded team within Google to work on this sort of thing? Yeah. So basically Google was kind of a first mover in thinking about how important understanding behavioral science was to product and marketing development. And so it was Dan and I in the beginning. And over the course of the three years, we hired internal Googlers. We brought in more people on our team and really kind of were an internal consultant where we'd [17:42] be pitched to other teams, work with them from start to end on a project. [17:46] And now Google has one of the bigger behavioral science teams that's out there. Wow. Okay, so... [17:52] You've worked with how many companies at this point, roughly? I'd say hundreds. I mean, yeah, hundreds. Okay, so there's hundreds of companies. There's also, I imagine, hundreds of these kind of biases and heuristics and kind of behavioral expectations of people. I'm curious, of the work that you've done, what are the most common biases, heuristics of people that you find most useful in improving product experiences and product design? Like, what's kind of the handful that ends up being most useful? [18:21] Totally. And there are so, so many mistakes that humans make and we can call them biases, heuristics. Our team uses psychologies. What are the psychologies that drive us? And so to tackle this, [18:31] Our team basically has created a model of behavior change we call the three B's. And this summarizes the most important psychologies that drive user [18:40] that are important to the product manager and the market artist. [18:43] We've used this at Google, Microsoft, LinkedIn. So all the companies that we work with and they now use it.

[18:49] and so the first b so 3b framework first b [18:52] It's actually not a psychology, but it is the most important part of behavior change. And it should kind of be obvious, like if you think about behavior, [19:00] change behavioral economics. It is [19:02] behavior. In order to change behavior, you have to pick a behavior that you want to change. So, [19:07] companies are really good at outcomes but just not as sharp at picking the behavior and when i say behavior i mean action the thing that you want someone to do by the way the only wrong answer here is log in [19:19] So it's really, it's not about logging in, it's about what you do after you log in. [19:24] And when we're consulting teams, it'd be like, [19:26] we need to get uncomfortably specific, we say, just really specific in the behavior. So, [19:31] Example, if I'm Peloton PM and I'm working on the app, [19:35] I would say something like within seven days, [19:38] of somebody starting the app they do two 10-minute workouts with two different instructors [19:44] Now obviously that is wildly specific and you'd probably be very happy if they did one workout with one instructor. [19:50] But the reality is if you don't define that behavior you're going to change, you can't actually define the psychologies. [19:56] that affects someone's decision making when doing that behavior. [20:00] So that's the first P. Second P, again, [20:03] is probably pretty obvious and is very critical is just barriers so we need to reduce the barriers to doing the behavior and there are two types of barriers we we look at one is logistical [20:14] So this is just the stuff in our way. It could be entering a credit card. It could be any form fields. [20:18] And then the second is cognitive. So the cognitive barriers get in our way as well. These are

[20:23] things like uncertainty aversion. [20:25] This is optimism bias, information aversion. [20:29] Right. These are the things that would prevent status quo. Big one. It's just you do the same thing today that you did yesterday. Your job is that you talk about those three just throughout there just briefly. I'm curious about the specific. Yeah. So uncertainty aversion when something is uncertain or we're not. So I'll give you an example. If you're Lyft. [20:49] There's logistical friction, which is wait time. [20:52] But then there's also this uncertainty of, is it going to come on time? When is it going to come? And with this uncertainty, you're probably going to look for other options. You're going to open up Uber and say like, [21:03] Maybe it'll come faster. [21:05] And so when there is uncertainty in our life, we either [21:08] look for other options or we just don't make a decision at all. [21:12] Right. [21:13] This is, by the way, very big in health care, where when you're very uncertain about something, you may not even go to the doctor. [21:20] or you may just make the wrong decision. Got it. And same with status quo effect where, you know, we kind of underlying status quo effect is this idea that we always take the path, not always, but majority of you don't really say always when you're talking about human behavior, human behavior is very complex, but more often than not, we take the path of least resistance. So we do the thing that's easiest. [21:40] And typically the thing that's easiest is the thing that we've done yesterday and the day before. [21:45] And so when you're asking someone to do something different, which is what most product, especially startups are trying to do. [21:51] You actually have to increase their motivation or make it easier, reduce the barriers to get them to do that. And so.

[21:58] Status quo, in fact, is a big... [22:00] big one that folks are fighting. Awesome. Thanks for sharing this. Let's keep going. Cool. So the third B is [22:08] benefits so this is where you want to increase [22:11] the not just benefits but the immediate benefits of doing something so we are all present bias which means we [22:18] Prioritize our present self over our future self. [22:21] So there are plenty of reasons [22:23] somebody your customer your user should take an action but you actually have to give them a reason to take an action today. [22:29] So as an example, if you're, [22:31] Asana and... [22:33] they're trying to get someone to log a task, the right thing for them to do is log the task because it's going to get their project done on time. [22:40] You're going to have a collaborative and communicative team that you're going to want to be on. [22:45] But one of the real reasons we may log a task is because of completion bias. We want to see the checkbox. [22:51] We may log it because of social desirability bias where [22:55] Other people see that we're getting our work done. There's a notification that goes to my teammate when I complete something. [23:01] So these are the immediate benefits that we have to build into products and features to drive use. Awesome. I definitely have completion bias. I love checking those freaking checkboxes. Yeah. I know Sana is famous for that little unicorn confetti thing, right? For like making you feel really good when you complete some. [23:17] Yeah. You know, I don't know how much emphasis to put on these kind of like [23:22] you know, confetti stuff. I think the idea of being done is that that psychological understanding that you've completed it.

[23:30] Confetti or not is very powerful. I love confetti. Okay, so just to recap, there's the three Bs, behavior. What is the behavior you're trying to change? And the way you described it sounds like the activation milestone is the way most companies think about that. Is that right? [23:46] Yeah, I would say the activation milestone and [23:50] The novel thing here is just getting really specific. [23:53] about this behavior so much that if you're not arguing with your teammates you're doing it wrong oh about which behavior and yeah i imagine this often comes similar to the activation kind of moment it comes back working backwards from what is most likely to keep this person around long term kind of to keep them retained right [24:11] Yeah, exactly. And yeah, so what correlates with retention and what provides kind of an immediate value to them by doing this behavior. Awesome. Okay, so behavior, then barriers. How do you... [24:21] Breakdown barriers, logistical and cognitive, and then benefits. And the point about being an immediate benefit is a really interesting one because, yeah, it may be like in a year from now, I might earn some money, but I'm not going to do anything hard right now if I'm not going to make any money right now. [24:36] Exactly. And so this is a tough one because we want to believe that people really do [24:41] and use our product for the end benefit, [24:44] but we really need to give people the reason to do it today. Awesome. I want to dive into a couple case studies of companies that I know you've worked with. Maybe before that, I thought it'd be good to ask this question about just like the dark side of this stuff. I imagine people often worry about, "Hey, are you just tricking people into doing things they don't want to do? That's not good." How do you think about that kind of line?

[25:06] Yeah, it's a great question. I think the answer to it's really [25:10] when many things in life comes down to incentives. [25:14] So what are the incentives of the product teams, of you, your manager, the team working on this? I like to say kind of we are what we measure. [25:23] and it really matters what you measure. So I'll give you a personal example. Lending Club, a while ago, hired me to increase their conversion flow for borrowers. So they're a, [25:34] Lending. [25:35] and they wanted to get more borrowers in the flow, [25:38] The novel thing here was that they would pay me a lot of money if I hit a bump. It was like a five-point bump. [25:45] and zero if I didn't hit anything. That was your incentive structure. Right. So my incentive structure, [25:52] was to increase the conversion flow. [25:54] And [25:55] Over the course of working with them, I became a predatory lender. I was suggesting things that the legal team was like, no, Kristen, of course you cannot do that. [26:08] in setting some context i was at that time leading [26:12] a duke university lab called common sense lab it was focused on low to moderate income americans thinking about their financial health building savings apps for them so here i am this kind of like do good non-profit very focused on financial health of americans and my other [26:30] Yeah, well, Predatory Lender for Lending Club, right? And so it's really about the incentives that we're giving ourselves and our team that will drive how anything is used, much less behavioral economics or these tactics, but generally what's the success of your team. You may not have a clear answer to this, but any advice for setting those incentives correctly? Anything you've seen work well?

[26:52] Yeah, I think for us, there's two tips. One would be to set the incentive on the behavior, because the behavior is going to align more with the customer outcomes than the active use or the retention. And of course, and I'm not suggesting we don't, [27:05] measure active use and retention and these things you should definitely measure them but setting the [27:10] incentive and the KPI and the product and the marketing team on the behavior is going to result in a more customer friendly [27:16] product and outcome. [27:18] And then second would be increasing the [27:21] duration of the incentive. So if you're trying to hit a quarterly outcome, [27:26] you're likely going to do things that will impact the short term [27:29] to give that a bump. If you're trying to hit an annual... [27:33] or even longer outcome, you're going to be looking out for the best interest of the company and the consumers in a little bit different way. Awesome. For the Lending Club example, what did you end up doing in the end, by the way? Did you just like, ma'am, let's not do this? I can't actually reveal the intervention that we did, but I did miss it. So there was a five point, by the way, point, not percentage. [27:51] bump, which is a lot, and we got 4.5. So we almost hit it. But [27:57] It's a clip, it doesn't work that way. - You almost sold your soul for that bump, and you didn't get there. - I know. - Seems like the right choice. - It was a good lesson in incentive design. And by the way, at that point, Lending Club was offering good rates to borrowers, so, [28:10] There was efficacy in the idea of it, which is, [28:13] lending clubs, lending to people who have lower FICO and with better rate. [28:17] And yet you still want to have, you know, you don't want to become the predatory lender in that situation. Okay, awesome.

[28:47] a better place to hire product managers and growth leaders. Join almost 100 other companies who are actively hiring through this collective. And if you're looking around for a new opportunity, actively or passively, [28:59] Join the collective. It's free, you can be anonymous, and you can even hide yourself from specific companies. You can also leave anytime, and you'll only hear from companies that you want to hear from. Check out Lenny's Jobs dot com slash talent. [29:15] So there's two companies that you work with that I use. And I'm curious what you did for them and how you worked with them. One is TikTok. And then the other is one medical. And so it'd be cool to just hear what you did for them, what they wanted your help with, and then just kind of what you took away, how it went, all that. [29:31] Great. Yeah. So TikTok is obviously a big platform and with any big social media platform, they have misinformation on it. So this is not just a problem with TikTok, Facebook and Twitter have it and misinformation is growing. [29:46] And so they asked us to help decrease the spread of misinformation on their platform. So there's lots of ways that we could take that. And the first thing that we did is we kind of do with any companies think about what is the behavior that you want to change. And so we narrowed. [30:02] on decreasing shares so we could have said likes you know we could have said comments or general engagement which you know is a broad thing but we narrowed on decreasing [30:11] the shares. So when you're actually going to see a misinformation video, [30:15] We want to decrease the number of people that share it.

[30:18] So that was the first step. And the second step is this, [30:21] you know, 3B model, how do we think about the barriers? [30:24] And so, [30:25] You know, with TikTok, [30:26] the thing you want to do is get people to do something less. When you want to get somebody to do something more, you make it easier. [30:33] When you want someone to do something less, you make it... Put up barriers. Put up barriers. Yeah, actually, a nice example of an elevator study where basically you want people to use the stairs... [30:44] And so what the researchers did was make the elevator door close 16 seconds longer. [30:50] So now you're getting the elevator, you're annoyed, it's harder to use the elevator. [30:54] and people went to the stairs. 16 seconds, interesting number. 16 seconds, we can't wait 16 seconds. That is actually kind of long though. You're standing there, it's holding your thumbs. I'd use the stairs. Yeah, yeah. And so that's kind of what we did for TikTok, which is basically you're in a very hot state. TikTok is a very fast platform. And so we just slowed people down. [31:12] So, [31:13] Two things, there was a label on the video, which is kind of classic in the misinformation world where you tell users that it's unverified information. And the second thing was once you click the share button, we popped up something that said, are you sure? [31:26] And this was basically slowing people down, so logistical friction, and actually calming the friction because we're asking people to reconsider a prior decision and question their self. And, [31:36] And so with that, [31:38] both those things the label on the video and the pop-up reduce shares by 24 and these are tactics that facebook and twitter are also using to slow people down this was actually one of the first though published studies that showed that this

[31:50] this drove down misinformation. Can you talk a bit about how you came to those two conclusions? I know you work really closely with the product team and there's this whole process you go through. How do you come up with these ideas? How many ideas did you test? Is there anything more you could share on that? Yeah, it's actually a pretty involved process. So first we did a literature review, which is basically saying like, look, we're not the first people to think about this. Let's [32:11] talk to the experts and understand what has worked and what hasn't. So with that, actually one of the insights was that reminding people about their value of accuracy, [32:21] So most people want to be accurate. Most people actually don't want to share misinformation. [32:26] And so if you can remind people the point of sharing, by the way, it doesn't work if you remind people before or after, like it's at the point of sharing that this is unverified information. Studies have shown that this could decrease share. So we took that piece. [32:39] We also took this, you know, like the hot and cold state literature that says if you actually don't want to get someone to do something, how do you intervene? So we did this literature review. We came up with a hypothesis and we had probably 30 different ways to implement this because the hypothesis is just a starting point. [32:54] And then we put it into quantitative research. So we actually put five different versions. [33:00] of the pop-up. [33:01] in front of over a thousand users, which we got users from Prolific, which is a platform, [33:08] where you can easily test things. [33:09] And we didn't measure the idea of like, [33:14] do people like this. We measured a condition against another condition. So relatively,

[33:20] what condition is more likely to work than the other condition and that would increase our intuition that it [33:26] would work in market. We're not saying that studies outside of the product are [33:31] end-all be-all, but it's going to drive our intuition up a lot more. For the research that you did, [33:36] I imagine that's a lot of the value that you and your team bring to a group. For folks that are just on their own, maybe, is there any advice for how to find research on problems you're trying to solve? [33:46] Yeah, I mean, I would say basically, if you're looking at a new problem, [33:50] It's very tempting to talk to five users in the beginning. We would basically say, look, go spend a day on the Internet. [33:57] Googling to see what else has been done. And Google Scholar is a great place to start. [34:02] Like hot tips would just be making sure you're searching the right keyword. So if you're looking at something in healthcare, you'd want to know what are the buzzwords for, [34:10] chronic care and things that would actually drive you to a meta-analysis that would summarize what's already been done. [34:16] So Google Scholar, great place to start. [34:19] And then... [34:20] Coming up with these hypotheses and putting them to test, we never do a UX study where we're just showing people one thing. [34:29] because they could really like it or hate it, but they could really like or hate all the designs. [34:34] We have no idea. [34:35] So when we're doing this, we always present multiple options and then relatively look for which one is going to drive the behavior we're intending to change. Is there a reason you did UX oriented test versus testing the product? Is it just because they weren't ready to build an experiment in product yet? Yeah, we actually got two conditions in product, which for us is we would like more. And it was, you know, we don't know if we're going to get a second bad at this. So we said, well, two conditions plus the control. But what should those two conditions do?

[35:05] be. [35:05] So for us, it was like we want to de-risk the launch as much as possible. Awesome. And is that feature in the product today? [35:13] It is. Yes, it is in the product. And yeah, they've done more global launches as well. Sweet. I am clearly not sharing much misinformation because I haven't seen it yet. [35:26] You may not be searching the right thing. So if you search medical COVID politics, you may get it. Oh boy, here becomes the decline of me. Okay. So next case study that I thought would be fun talking about is One Medical, [35:43] Thank you. [35:43] Yeah. So we worked with one medical to [35:47] Basically, one medical ask is to increase engagement. This is a classic question. How do you increase engagement in a product or service? Engagement being just using it more often. [35:56] Right, exactly. But that's the real question is what does engagement mean? Actually, a fun exercise that people could do with their teams is if you ask everyone on the team what engagement means separately, [36:06] you can compare answers and most likely people are saying different things. And if you guys can't agree internally on [36:13] what to do or what engagement means. It's very difficult to build a product by which your users understand what to do. Yep. Yeah. So after lots of debate, we came up with a key behavior to target basically somebody setting up a doctor's appointment. [36:28] right after signing up for One Medical. And this is really important because [36:33] more likely than not you're going to sign up for one medical six months later you're going to get a fever and get sick

[36:39] And are you going to remember this app that you signed up for six months ago? [36:43] Probably not. You're going to do the same thing you did before, which is call the doctor that you've normally gone to. So really important for one medical upfront during onboarding to set the mental model of what they can be used for in going to that doctor's appointment. [36:58] So then we looked at the barriers, one of the barriers, like who is going to be your provider, your doctor, like picking somebody is hard. Finding a time, an appointment time is hard. You know, we're all busy people. How are you going to fit it in? [37:11] Knowing actually what One Medical can help you with. [37:14] also difficult like [37:15] What what is this company or thing and where should I what should I go to them for and then the benefits, right? What are the benefits and the benefits? [37:23] For using one medicals, you get [37:25] Your question is answered immediately. [37:27] it, [37:27] very simple but you can really get your health questions and issues addressed. And so the simple intervention we did was just during onboarding, [37:35] We ask people a few questions about their health. [37:37] mental, sleep, physical. [37:40] We then recommended a provider. So instead of having lots of choice, decrease the choice, recommended them one provider. [37:47] Tell them why. [37:48] and then recommended appointment times and limited the amount appointment times it was for tomorrow we recommended a virtual appointment so it would be a quick [37:57] thing that you could do from your home. [37:58] And then that intervention, like the onboarding intervention, increased [38:03] bookings by 20% during onboarding. Of appointments. Of appointments. Yep. Yeah. [38:09] Onboarding.

[38:10] I don't know if I went through that when I joined One Medical. This is recent, yep. Okay, cool. Yeah. By the way, One Medical is awesome. As you're talking, I'm thinking just if you could think about what are like the five most commonly successful changes to a flow? [38:26] One right now you just shared was reducing options, which is interesting. And then another you mentioned earlier, just like reminding you of maybe what you're going to get out of at the end. I don't know. Is there anything that comes to mind of like, here's like a subset of things that seems to often work when you're trying to improve, say, activation or conversion of a flow? [38:56] course there are [38:57] There are some rules of thumb. [38:59] So one thing really is about showing people the benefits [39:04] right away. [39:05] So really not under or overestimating [39:09] that people understand the benefit of your app. And so how do you get them to experience it? [39:14] immediately. [39:15] So that's one. And then the other is really just the reducing [39:20] I hate to sound so simplistic, but this is, you know, a lot of it is about making things easier logistically and cognitively and making, [39:28] the benefit much more clear salient we say moving it from abstract to concrete [39:34] Like you really want to move something [39:36] you know from the world of people can't really imagine how it fit in their life to how does it actually fit

[39:43] in their lives and and keeping them going you know it's [39:46] if it's a long sign up flow and people need to drop out how do you send them reminders to pull them back in and then when they are back in land them on the spot that they were at so wealthfront has said [39:58] you know, one of their key insights was when people dropped out of the flow, [40:02] Normally, they would just put people back at the beginning, [40:05] It's really frustrating if you can put people back where they landed, they're more likely to continue. [40:11] That makes sense. Reduces barriers. I'm learning. So there's two really interesting things you just said that I'd love to dive into a little bit. One is showing the benefit of something right away. Is there anything, is there an example that comes to mind where you did that and it worked out really well? Like what's an example of showing a benefit? [40:27] Yeah. So actually benefits can be, there's two ways to do benefits. One is the actual one that [40:33] a consumer has so right asana and the like collaborative communicator versus [40:39] maybe one that's more immediate. And so when we're thinking about more immediate benefits, [40:44] This could actually be things like social norms, where you're telling people everybody else is doing something. And to them, the benefit of doing this is following the herd. We all, by the way, don't like to think that we're part of the herd, but we all will succumb to this at some point in our lives. And so, you know, one example is where we worked with a company called. [41:04] study, [41:05] and [41:06] it was a fintech app and we're helping people to sign up or and sync their bank account. Syncing their bank account is a required step in any kind of

[41:14] FinTech app and yet so difficult to [41:17] Big drop off point. [41:18] And just by reframing, [41:20] the choices of getting someone to think about completing a step, [41:24] you know, versus the benefit that would come to them. [41:28] after. So the benefit of sinking their bank account was actually huge. You'd get to see your income, [41:33] You're really actually changing someone's financial picture by doing it. And they're signing up for it. They know it. [41:38] But the small immediate benefit is how you're framing each step. [41:42] Each step is going to give you the immediate benefit. So for this one, it was a completion bias thing again, where people are motivated to complete something. Sometimes we, you know, we'll do an error message if it's really important for the app to work. [41:55] And again, that's not you shouldn't do that all the time. But if it's really important, the benefit is actually decreasing that error message on the screen. We call this right for wrong doing the right thing for the wrong reason. [42:05] where the right thing may be completing the flow, but the wrong reason is seeing that error message go away. [42:10] Again, I would use that with caution. People will not use and love your app if you're putting error messages all over it. [42:16] That makes sense. I think you mentioned this. So I haven't mentioned, but you wrote a guest post for my newsletter. It's one of the most read posts of the newsletter on using a lot of the. Yeah. I don't know if I told you that. Yeah. And it's about using these insights to improve conversion in general. So we'll link to that in the show notes. And I think in that post, you mentioned this concept of right for wrong. It might be worth spending a little more time on this. Just like what does that mean? And what's what's an example that when we think about right for wrong, we're helping people again, do the right for wrong.

[42:46] thing. [42:47] So this is a behavior that they want to do, but the reason is actually [42:51] probably not related. So if you think about just [42:54] you know, vaccines and what we saw here, like, [42:56] The reason to get a vaccine is to help my neighbor and prevent them from getting sick. [43:01] And yet, [43:03] What happened? We gave people donuts, give people free pot or marijuana. There were lottery tickets in Ohio. So this is the right for wrong idea where you're really substituting. By the way, there's one of the bigger voting interventions is just, [43:18] having pizza show up at the poll line, right? You're motivating people to go vote not because they want to contribute to democracy but because there's pizza [43:25] and the poll line so at some level we're pretty simple human where we want an immediate [43:30] benefit. [43:31] Um, [43:32] Clubhouse actually was an interesting example of this at some level where [43:37] you saw two levels of status in Clubhouse. There were the people who were friends with the people on stage and then there was everyone else. [43:44] And so, [43:45] That was like a motivator to be friends with more people and start following people so that you could be in that little bucket of friends with other people. [43:56] So. [43:56] we're giving people kind of some other reasons to do a thing that, [44:01] they may want to do but they don't necessarily have the motivation to do it right away. [44:06] If you think about Peloton, you know, [44:09] Yes, I want to work out. Yes, I want to have a healthy person. [44:13] but maybe I want to go to the class because the instructor may give me a shout out.

[44:18] or [44:18] there is some streak that I'm trying to continue. [44:21] So they're right for wrong here. It's like the real reason to work out is to improve my physical health. [44:26] But wow, I just want that instructor to give me a shout out. So the key there, just to kind of come back to the dark side of some of the stuff, is you want to make sure the end mean is good and right. Exactly. Yeah. And so again, all these things, you know, the possibility of using anything for negative outcomes exists. And really it's about the incentives is what is your incentive? [44:48] to drive the customer benefit and that's really where we say you have to be measuring the customer [44:54] outcomes. You have to be focused on the customer outcomes. And some of the toughest teams we have understanding this are nonprofits. So we worked with one company a while ago, we tried to get them to add a deadline because they were trying to help people sign up for interviews. So like, [45:11] They were basically filtering people before Staples would get them for an interview. [45:15] They'd do the interview prep, they'd help them show up for the interview. But the reality is that sometimes people miss this stuff, it takes a long time to get to the interview. [45:23] And so our push to them was to add a deadline. [45:26] to help people show up. Sometimes you say deadlines are a gift, right? You're just helping people kind of prioritize this. [45:33] And they said, no, absolutely not. We cannot add a deadline. [45:36] This is just too hurtful. And the reality is when you add deadlines, and we've done this in multiple companies, where you add a deadline, more people do the thing. [45:45] right? Because you're helping kind of create the right for wrong reason, giving people the wrong reason to do the behavior that they actually want to do. We added a deadline to Kiva flow for when their borrowers were signing up and we increased their conversion.

[46:00] So, [46:01] Obviously, you can use this in ways that are unintended, and it really has some upside to help people. I imagine you're building kind of this just like toolbox of ideas that end up working often at Flow. I know you said that you can't just copy and paste. Rarely is it just like, oh, this worked at 10 other companies. It'll probably work here. But that's kind of a cool thing that your team is probably doing, just making this big list of like, here's 10, 20 things that we could start with these ideas and talk about them. [46:31] to understand and see patterns. [46:33] We started with the idea that behavioral economics shows that people act in irrational ways, but we do it in predictable ways. [46:40] And so once you start understanding this, you can really start designing for the behavior change at this kind of predictable level. For a product team that's listening to this and they're like, oh, we we can't work with you guys right now. We we can't afford it. We don't have time just like generally. [46:56] but I want to think about this stuff and maybe implement it on my team. [47:00] What would you suggest to, say, a PM leading a team that wants to implement some of these principles? [47:06] Yeah, I would probably start from the process angle. You know, we have the 3B framework, which, again, the most important step is defining the behavior. So if you were to do a workshop, [47:15] with your team on what is the uncomfortably specific behavior that you want to change [47:20] and you align [47:22] everyone to this, it's [47:24] going to be much easier for you to change that behavior because the psychology that drive that behavior will be [47:29] a better conversation and more obvious. So first, starting with the behavior,

[47:33] to get everyone in a room, do a small workshop. The second would be that [47:37] Behavioral diagnosis and this is the toolkit in behavioral science where we study not what people say they will do but what they actually do. [47:46] And so the behavioral diagnosis, it's like a journey map on steroids. [47:49] where you're really trying to map the steps that get people to the behavior change. And again, just doing this, teams have light bulbs. And this is one of the main things that we work with teams on. We help them do the behavioral diagnosis to have those light bulbs. [48:03] And then we help diagnose and design the intervention. Can you talk a bit more about this behavioral diagnosis step? Like what actually happens there? [48:11] There's kind of two steps. One, you can outline the full kind of high-level flow where you're saying, you know, maybe there's... [48:18] 50 steps in getting to the behavior and you can do this before they hit your product too but when we're working with just like a very kind of we want to help a team grow their product versus develop a new product it's just about growing [48:32] and improving the current product. [48:33] We'll do a deck of... [48:35] 200, 300 slides of screenshots. [48:37] where it's just... [48:39] very detailed analysis of the steps that it takes to get to the behavior change. And then we overlay the psychologies, [48:47] that are driving people at each step, right? This actually takes a really long time. [48:52] But the light bulbs are like how you actually get to that. As I said, behavior change isn't easy. You know, it's complex, it's hard, it's noisy. And so you have to understand the steps that get you there before you're trying to change it.

[49:04] Wow, that sounds really useful. So you go through the product, make a screenshot of every little step and then connect what barriers keeping them from progressing. Is that roughly right? Exactly. Yeah. So maybe in a health app, it's information aversion. They don't actually want to see their test result. That's a step. [49:23] And then, [49:23] we attach that psychology to that step. And now we can have a brainstorm about, well, what would help decrease information aversion to get them over the next time? What's cool about this is like you could just, as a team, you have all these things you're trying to improve, all these flows and steps and things like that. And basically these psychologies, as you call them, are useful to just like anything you're trying to optimize. I imagine there's just like a list on Wikipedia of all these human psychologies and biases. I imagine a better [49:53] a good time to just chat about the course that you teach versus coming in-house and working with the team. Can you talk a bit about that? Or are those the same thing? Maybe I'm not sure. No, that's perfect. Yeah. So we have a... Thanks for the tee up, Lenny. Appreciate it. We have a bootcamp course, [50:10] which basically is a self-paced online course that you can go in and understand. And that's a lot of our case studies, many of them I didn't talk about here. And we walk you through the process, right? And you'll have a initial call with us. There's a Slack group, there's office hours, [50:29] really trying to help people go through the process of behavioral design [50:32] yourself and it it really brings these concepts to life and helps you apply them to your to your product so

[50:39] I think starting with that bootcamp is a great [50:42] place for designers, product managers, marketers, researchers, [50:46] to start we also we can i can send you this we have a sweet 16 like some of the 16 biases that we you know that we refer to most the three b's is an easy way to evaluate a product if you're just starting but as you get deeper [51:00] these different psychologies will pop up. Awesome. Please send that and we will link to it in the show notes. Just a couple more details on this course. How often do you run it? How long does it take generally for folks to take? Yeah, so once a year we'll do a cohort course that's a little bit more in-depth, but the best way is this kind of self-paced course. It's an eight-week [51:19] course. [51:20] and we give you a month extra to go through it. There's 16 modules. There's homework throughout it that you can post in Slack and get feedback from behavioral scientists. So a pretty kind of interactive course, but you can go through it on your own and even share with your team. We do group things. So we'll do like a team of [51:40] 10 and then we can give you a private office hour where we just riff with you on your product challenges. Awesome. And folks can find that at irrationallabs.com. Is that right? Yeah. Thanks again for this. Yeah. Arrationallabs.com is the spot. And if you're probably signing up for a newsletter, it's also a move we publish all of our research there and other kind of fun behavioral insights. Sweet. Well, with that, we've reached our exciting lightning round. So I'm just going to ask you a bunch of quick questions, whatever comes to mind. Fire away and we'll just go through them.

[52:10] fast. Sound good? Great. Okay, cool. What are two or three books that you recommend most to other people? I would be remiss not to [52:18] say particularly irrational i think this is a great starter book for people getting interested in the field [52:23] Influence is also a great starter book for people getting interested in the field. Many times you read Influence and you change your career. [52:30] It's a really powerful, powerful book. And then an under talked about one is I love the author Robert Frank, who wrote Darwin Economy. He talks more about kind of status and social norms and how this drives a lot of our behavior. So I'd recommend Robert Frank as an author and Darwin Economy. What's another favorite podcast other than the one you're currently on? [52:51] Science of Change is my podcast, so a little self-promotion. We interview product leads and talk about what they've done in their product. And then the other one is No Stupid Questions. So this is [53:04] by the Freakonomics, Steven Debner and Angela Duckworth come together and just answer questions about behavior change. What's a favorite recent movie or TV show? [53:13] So I love the rehearsal. This is, yeah, so it's basically where you, Nathan Fielder is rehearsing life. [53:21] And I think, you know, [53:22] Sports is so obvious. You want to practice. I just started ping pong lessons actually like I'm doing the same stroke for the net for an hour. [53:29] But like in life, we don't rehearse as much like. [53:33] you know, I don't write an email 100 times and then practice sending it. And so I think there's probably an opportunity for us all to learn some lessons from that. Have you seen Nathan For You, his other show? I have. Yeah, the rehearsal was just so compelling. And I'm not usually somebody who watches media. So this was like, mind blowing. I loved it. I actually like Nathan For You better. Oh, interesting. Okay, I'll give it another. What a weird show, but I love it.

[54:03] So interviews for, I'll take this as interviews for, [54:06] people when we're hiring. [54:07] I would caveat to say, so interviews actually we know don't predict [54:11] job performance. So, [54:13] Very little to no evidence and say interviews will predict how I perform on the job. And so yeah So we use you know skilled assessments and trials as ways to predict job performance and [54:25] But interviews are good at predicting affinity and culture fit. And so if you want to ask a question about that, [54:31] you know one of my favorite questions is just what is a personality trait that defines you so really [54:38] one that you feel the strength, [54:40] but that also can be seen as something you're working to improve. [54:45] So for me, I may be confident, but that may mean that I'm not listening. [54:50] And so we have really strong personality. People are really quick at answering this question. They really know the thing that they do good, but also could hurt them. And I find that it helps understand a person's [55:02] a little bit easier and faster. Last question, who else in the industry do you most respect as a thought leader? So I don't know if people will [55:10] know him but Chris York so he [55:12] is a behavioral scientist, [55:15] designer just really just know i would go to chris and and ask him a question that he has [55:22] done full research, just accidentally in the last year and done full research on the question, has a very thoughtful answer. And yeah, Chris York wins for me. Amazing. Have not heard of him. We'll look him up. Kristen, thank you so much for being here. I feel like you're going to change a lot of people's behaviors. You're going to break barriers.

[55:40] for a lot of products, and there'll be many benefits that come out of our chat. Thank you again for doing this. Wonderful. Thanks for having me, Lenny. This was super fun. [56:10] .