The Exchanges

Q Buy the freeze, that's why they're there. But okay, so lessons on how to do that effectively, is it just create addictive apps that people don't want to?

A Well, like, I think the core experience of course matters a lot. Addictive sounds like the bad version of this, but habit building, like you want to build good habits. And the reason we use that term is like, if you want to learn the language, you have to practice, if you want to be fluent, thousands of hours. And to get to thousands of hours, you need a daily habit. You can't like binge your way to fluency. So that's why everything at Duolingo is built around a daily habit. Once you've done an experience that people want to use daily, you have good daily retention. Then you kind of layer on these benefits that are kind of extra like streak and a streak freeze. You don't need it, but it is really core to your daily habit and then freezes this extra benefit that you can monetize. That's how we've thought about it.

AI assessment note: “Addictive sounds like the bad version of this, but habit building”

Q I do have to ask, because you mentioned at the beginning, like, being central to the monetization. And I spoke to Severin, and he said that, bluntly, it was your fault that they waited so long. Uh, jokes aside, why did it take so long, do you think? And how do you reflect on that decision?

A The hindsight bias is easy to apply to our past and say, oh, we waited too long. But I think at the time, a few things were I would say maybe headwinds to starting to monetize. One of them is Duolingo obviously has a very strong mission today, and that certainly was the case then. When it first came out, Luis, you can find Luis's talk. This was a lot easier to find on YouTube, and I think it was on our corporate pages. I think we removed it now because we moved away from that, but the monetization plan for Duolingo was going to be, everything was going to be free. So as users went through, did language exercises, they would do translation and then We would crowdsource those translations as a service to anyone who wanted to buy it. So if CNN wanted to translate an article from English to Spanish, they would feed into Duolingo system. That article would be broken up into chunks, fed into users and exercise. And as crowds translated that as an exercise, it would populate back a translation. That was the original plan. And as Luis announced Duolingo, he also said, we will never have ads. Never in-app purchases and never subscriptions, and finally left us, those are the exact three things I did as part of my job at Duolingo.

AI assessment note: “I think at the time, a few things were I would say maybe headwinds”

Q How do you think about the right time to hit them with a paywall? You want to build the trust, the love, the loyalty without being like, okay, level one paywall.

A Exactly. So I think it's twofold. One of them is honestly like hitting with a paywall is kind of the wrong approach in the sense that I think you want people to discover the paywall by experiencing features. That is the best way to honestly like convince someone to buy something. Meaning let's say I would like to convince our users to buy Duolingo Max and it's It's key feature is a video call. The best way to do it is have them try video call actually. So a lot of our ways to do that is maybe show them a video of how video call works and then say, if you're interested, here is Duolingo Max or, uh, have them go into the experience, have a call themselves. And then at the end say, this is part of Duolingo Max. That is kind of the best way when it's like contextual and users discover that there is something to be paid to like experience. I think just like having people showing kind of what I would call like generic ads works way less better. And it looks spammy in the app. So we try to do everything contextual. For example, if we show you an ad, we will say right after there's an ad free version if you'd like to upgrade. And that is in a way like you've discovered an ad. Okay. Maybe you didn't love it. Okay. But you can also find an ad free upgrade.

AI assessment note: “you want people to discover the paywall by experiencing features”

Q Buy the freeze, that's why they're there. But okay, so lessons on how to do that effectively, is it just create addictive apps that people don't want to?

A Well, like, I think the core experience of course matters a lot. Addictive sounds like the bad version of this, but habit building, like you want to build good habits. And the reason we use that term is like, if you want to learn the language, you have to practice, if you want to be fluent, thousands of hours. And to get to thousands of hours, you need a daily habit. You can't like binge your way to fluency. So that's why everything at Duolingo is built around a daily habit. Once you've done an experience that people want to use daily, you have good daily retention. Then you kind of layer on these benefits that are kind of extra like streak and a streak freeze. You don't need it, but it is really core to your daily habit and then freezes this extra benefit that you can monetize. That's how we've thought about it.

AI assessment note: “Addictive sounds like the bad version of this, but habit building”

Q I was gonna ask, what did you do that you wish you hadn't done? A school's dashboard?

A Yeah, there's another one. I think that at the VC level, this makes sense, but when you get into the product level, it doesn't, which is, Why is Duolingo not catering to language teaching in class, in the classroom? Like, why does Duolingo not have a product that fits into the classroom? So we wanted to solve that. So we tried to come up with a version of our product that teachers could assign, teach certain parts of a curriculum and assign homework through the Duolingo interface, and then they could track if assignments are being done, et cetera. The problem with this is every language teacher wants to teach a different way, and the Duolingo Curriculum does not fit any of it, so you end up having to customize how people should go through Duolingo, and that's not the product we have. So even if teachers could assign assignments, that's not, that doesn't match the textbooks they're working on in the French classroom. Anyway, it didn't work.

AI assessment note: “we tried to come up with a version of our product that teachers could assign”

Q Was there a meeting? Can you take me to that moment?

A It probably wasn't one meeting, but there was a few key moments where I remember being like, ah, how am I going to figure this out? One of them was, uh, sitting with Luis in honestly, probably one of my first weeks, and I'm a very early product manager trying to figure stuff out. And I'm like, all right, Luis, like, here's our ads roadmap. I calculated how much money this is going to bake. Here are the experiments I'm going to run. And he's like, all right, what is it going to look like? And I show him like the prototype or like, we look at it. He's been looking at it. He's like, these ads look like shit. Shit is his first response. He's like, I'm not sure if we should do it. And I'm like, okay, uh, where did we, I, I hear you. I understand that this doesn't look like the pristine, every other page of the app. It's different. It's another advertiser's property. But we can improve the look. We can change the design. Anyway, it was an iterative process from there to be like, he's like, let's test. Let's see what happens. And the other worry was like, oh, this is going to kill retention. We test it. It's fine. It's like neutral on retention on the first experiment. And we're like, okay, our users are not running away. We are improving how the ads look and when they're surfaced. And then I think over time it started snowballing to be like, okay, we can do this. We can do ads. Once …

AI assessment note: “One of them was, uh, sitting with Luis in honestly, probably one of my first weeks”

Q Wow. That's fascinating. Okay. So that's like timing. What emoji usage you said?

A Yeah. Well, emoji breaks up pattern is, is why I think it works. So a lot of notifications suffer from They fall into a pattern, and now a sea of everyone sends so many notifications that, like, it all blends in. So you want to do something that is more interesting than the 85 other apps sending you notifications a day. Emoji is one way to break pattern. The other thing we do is kind of go full unhinged on notifications, so we write them from Duo's voice, and we make it sound like Duo is basically screaming at you, or to do your, not always screaming sometimes, sending you a subtle message, but it's like a personalized message rather than Duolingo saying, do your Spanish lesson or practice now, that kind of doesn't go very far, but if Duo says, yo, what's up? Are you doing Spanish today? And it looks like almost a message from Duo, that triggers you to be like, okay, yeah, I'll respond to Duo rather than this app that is sending me sanitized notifications.

AI assessment note: “emoji breaks up pattern is, is why I think it works.”

Q When you look at that process, and you look at the team that went into it, what does the structure of the product team look like today, and how will that change over time?

A This is a great question. I think today, the traditional product team is, you're gonna have somewhere between five to 15 people. That's the traditional product team that has engineers, designers, and product managers at its core, and it might have other functions like data science, marketing, depending on what they're doing. What I expect is going to happen, which is maybe the chess example is a good example, is these teams are going to be smaller because you might not need as many people to do the task at hand, which might be just like, we just need to create a prototype of this thing. We just need to figure out if this is the right design. We're probably not going to need as many people. So we'll have smaller number of people on each team. But honestly, I'm hoping by doing that, we'll be working on much more innovative bets quicker because again, like this chess example is a good one. If we can do that, like, 10 x more, well, Duolingo will hopefully be 10 x better over time.

AI assessment note: “today, the traditional product team is, you're gonna have somewhere between five to 15 people”

Q Was there a meeting? Can you take me to that moment?

A It probably wasn't one meeting, but there was a few key moments where I remember being like, ah, how am I going to figure this out? One of them was, uh, sitting with Luis in honestly, probably one of my first weeks, and I'm a very early product manager trying to figure stuff out. And I'm like, all right, Luis, like, here's our ads roadmap. I calculated how much money this is going to bake. Here are the experiments I'm going to run. And he's like, all right, what is it going to look like? And I show him like the prototype or like, we look at it. He's been looking at it. He's like, these ads look like shit. Shit is his first response. He's like, I'm not sure if we should do it. And I'm like, okay, uh, where did we, I, I hear you. I understand that this doesn't look like the pristine, every other page of the app. It's different. It's another advertiser's property. But we can improve the look. We can change the design. Anyway, it was an iterative process from there to be like, he's like, let's test. Let's see what happens. And the other worry was like, oh, this is going to kill retention. We test it. It's fine. It's like neutral on retention on the first experiment. And we're like, okay, our users are not running away. We are improving how the ads look and when they're surfaced. And then I think over time it started snowballing to be like, okay, we can do this. We can do ads. Once …

AI assessment note: “It probably wasn't one meeting, but there was a few key moments”

Q But I'm actually the founder of this idea and now company, because that was my idea. Uh, so thank you, Lewis. I'll take over. Um, would you say that you're an education company, an entertainment company, a content company? I know it's all Semantics.

A Honestly, neither is the reality. We have maybe in somewhat of an identity crisis, uh, is, is also the reality. We don't identify as an education company, but we also don't identify as a game or entertainment. So we're kind of in our weird own league and own market that we've created. The reality is what we compete for. If you're like, okay, but what are you? Who do you compete with? We eventually compete for screen time. Um, we would like to make your screen time Very useful for you to learn something and get better. The reason we don't fit into education very much is if you look at the education space, honestly, this is a bit of a hot take, but it's all boring ass products. So I think we would like to not be a boring ass product. As you can see, we try to do everything super fun. And I think if you say, well, you guys do a lot of fun, then are you a game? Well, we're not a game either. We also want to teach you something. So we're kind of in this zone league of our own is how we see it. But what we eventually compete for is your screen time. So everything that takes up your screen time, that might be social media, that might be gaming, that might be streaming, whatever it is, at the end is what Duolingo competes for.

AI assessment note: “We don't identify as an education company, but we also don't identify as a game”

Q What have you changed your mind on in the last 12 months? An example, for me, I was like, you know what, chat GPT, whatever, fuck it, we'll go to Grop, perplexity, you name it. I don't think there's any lock-in there. I'm like unwaveringly ChatGPT focused now. I have memory. I love the brand.

A Similar to you, I've actually changed my mind about the same thing, but I'll, I'll pick a different one. I thought this change in how we build product was maybe five to 10 years out. 12 months ago, if you had asked me like, oh, like, you know, how, how do you think the PM job will change? I'm like, well, maybe in 10 years we'll be using more AI tools. I think it's, it will get there slowly would be my answer. Honestly, now, probably in 12 to 24 months, the job of a PM will look entirely different. So I think we're talking about five years because that feels safe, but honestly, my guess now, or what I've changed my mind on is, it's going to be in two years, is my guess.

AI assessment note: “what I've changed my mind on is, it's going to be in two years”

Q And so I then say, you know what? Fuck this. I want no ads. And you're like, hmm, what happens then?

A A lot of users obviously were like, I love Duolingo, but you know, I'll pay you guys whatever. Can I remove the ads? That no surprise because that's a common model. That was the idea for our subscription. So it's like we had ads, we had a no ad subscription. That was the start. And what we saw also at the time was there was a lot of user love for Duolingo. It sounds like you also love Duolingo. And, uh, we saw users just buying the subscription because a lot of our users was, I was using Duolingo for five years. I just wanted to contribute something back, so I bought the subscription, even though I didn't know what was really in it.

AI assessment note: “That was the idea for our subscription. So it's like we had ads”

Q I was gonna ask, what did you do that you wish you hadn't done? A school's dashboard?

A Yeah, there's another one. I think that at the VC level, this makes sense, but when you get into the product level, it doesn't, which is, Why is Duolingo not catering to language teaching in class, in the classroom? Like, why does Duolingo not have a product that fits into the classroom? So we wanted to solve that. So we tried to come up with a version of our product that teachers could assign, teach certain parts of a curriculum and assign homework through the Duolingo interface, and then they could track if assignments are being done, et cetera. The problem with this is every language teacher wants to teach a different way, and the Duolingo Curriculum does not fit any of it, so you end up having to customize how people should go through Duolingo, and that's not the product we have. So even if teachers could assign assignments, that's not, that doesn't match the textbooks they're working on in the French classroom. Anyway, it didn't work.

AI assessment note: “Anyway, it didn't work.”

Q And so I then say, you know what? Fuck this. I want no ads. And you're like, hmm, what happens then?

A A lot of users obviously were like, I love Duolingo, but you know, I'll pay you guys whatever. Can I remove the ads? That no surprise because that's a common model. That was the idea for our subscription. So it's like we had ads, we had a no ad subscription. That was the start. And what we saw also at the time was there was a lot of user love for Duolingo. It sounds like you also love Duolingo. And, uh, we saw users just buying the subscription because a lot of our users was, I was using Duolingo for five years. I just wanted to contribute something back, so I bought the subscription, even though I didn't know what was really in it.

AI assessment note: “That was the idea for our subscription. So it's like we had ads, we had a no ad subscription.”

Q Why? Because everyone says that on the show.

A They love, they love saying that. First of all, the CEO is the CEO of the product, not the product manager. The product manager has to understand what the user will respond to when you kind of give them what they're going to see in the app. So they have to understand the user design really well. They also have to be able to stretch into business strategy, et cetera. That said, this is why consumer products live and die in the pixels and attention to detail matters. If a button that's like the dark shade of green versus the light shade of green will make a difference in user behavior, a product manager or a product designer for that record, like people working on the product should understand what that means. And I think this is also how we build product teams at Duolingo where everyone should understand pixel perfect design and why certain Specifically what the user is going to see, how that's going to change user behavior.

AI assessment note: “First of all, the CEO is the CEO of the product, not the product manager.”

Q When you look at that process, and you look at the team that went into it, what does the structure of the product team look like today, and how will that change over time?

A This is a great question. I think today, the traditional product team is, you're gonna have somewhere between five to 15 people. That's the traditional product team that has engineers, designers, and product managers at its core, and it might have other functions like data science, marketing, depending on what they're doing. What I expect is going to happen, which is maybe the chess example is a good example, is these teams are going to be smaller because you might not need as many people to do the task at hand, which might be just like, we just need to create a prototype of this thing. We just need to figure out if this is the right design. We're probably not going to need as many people. So we'll have smaller number of people on each team. But honestly, I'm hoping by doing that, we'll be working on much more innovative bets quicker because again, like this chess example is a good one. If we can do that, like, 10 x more, well, Duolingo will hopefully be 10 x better over time.

AI assessment note: “today, the traditional product team is, you're gonna have somewhere between five to 15”

Q But I'm actually the founder of this idea and now company, because that was my idea. Uh, so thank you, Lewis. I'll take over. Um, would you say that you're an education company, an entertainment company, a content company? I know it's all Semantics.

A Honestly, neither is the reality. We have maybe in somewhat of an identity crisis, uh, is, is also the reality. We don't identify as an education company, but we also don't identify as a game or entertainment. So we're kind of in our weird own league and own market that we've created. The reality is what we compete for. If you're like, okay, but what are you? Who do you compete with? We eventually compete for screen time. Um, we would like to make your screen time Very useful for you to learn something and get better. The reason we don't fit into education very much is if you look at the education space, honestly, this is a bit of a hot take, but it's all boring ass products. So I think we would like to not be a boring ass product. As you can see, we try to do everything super fun. And I think if you say, well, you guys do a lot of fun, then are you a game? Well, we're not a game either. We also want to teach you something. So we're kind of in this zone league of our own is how we see it. But what we eventually compete for is your screen time. So everything that takes up your screen time, that might be social media, that might be gaming, that might be streaming, whatever it is, at the end is what Duolingo competes for.

AI assessment note: “Honestly, neither is the reality. We don't identify as an education company”

Q What have you changed your mind on in the last 12 months? An example, for me, I was like, you know what, chat GPT, whatever, fuck it, we'll go to Grop, perplexity, you name it. I don't think there's any lock-in there. I'm like unwaveringly ChatGPT focused now. I have memory. I love the brand.

A Similar to you, I've actually changed my mind about the same thing, but I'll, I'll pick a different one. I thought this change in how we build product was maybe five to 10 years out. 12 months ago, if you had asked me like, oh, like, you know, how, how do you think the PM job will change? I'm like, well, maybe in 10 years we'll be using more AI tools. I think it's, it will get there slowly would be my answer. Honestly, now, probably in 12 to 24 months, the job of a PM will look entirely different. So I think we're talking about five years because that feels safe, but honestly, my guess now, or what I've changed my mind on is, it's going to be in two years, is my guess.

AI assessment note: “I thought this change in how we build product was maybe five to 10 years out.”

Q Wow. That's fascinating. Okay. So that's like timing. What emoji usage you said?

A Yeah. Well, emoji breaks up pattern is, is why I think it works. So a lot of notifications suffer from They fall into a pattern, and now a sea of everyone sends so many notifications that, like, it all blends in. So you want to do something that is more interesting than the 85 other apps sending you notifications a day. Emoji is one way to break pattern. The other thing we do is kind of go full unhinged on notifications, so we write them from Duo's voice, and we make it sound like Duo is basically screaming at you, or to do your, not always screaming sometimes, sending you a subtle message, but it's like a personalized message rather than Duolingo saying, do your Spanish lesson or practice now, that kind of doesn't go very far, but if Duo says, yo, what's up? Are you doing Spanish today? And it looks like almost a message from Duo, that triggers you to be like, okay, yeah, I'll respond to Duo rather than this app that is sending me sanitized notifications.

AI assessment note: “emoji breaks up pattern is, is why I think it works.”

Q I do have to ask, because you mentioned at the beginning, like, being central to the monetization. And I spoke to Severin, and he said that, bluntly, it was your fault that they waited so long. Uh, jokes aside, why did it take so long, do you think? And how do you reflect on that decision?

A The hindsight bias is easy to apply to our past and say, oh, we waited too long. But I think at the time, a few things were I would say maybe headwinds to starting to monetize. One of them is Duolingo obviously has a very strong mission today, and that certainly was the case then. When it first came out, Luis, you can find Luis's talk. This was a lot easier to find on YouTube, and I think it was on our corporate pages. I think we removed it now because we moved away from that, but the monetization plan for Duolingo was going to be, everything was going to be free. So as users went through, did language exercises, they would do translation and then We would crowdsource those translations as a service to anyone who wanted to buy it. So if CNN wanted to translate an article from English to Spanish, they would feed into Duolingo system. That article would be broken up into chunks, fed into users and exercise. And as crowds translated that as an exercise, it would populate back a translation. That was the original plan. And as Luis announced Duolingo, he also said, we will never have ads. Never in-app purchases and never subscriptions, and finally left us, those are the exact three things I did as part of my job at Duolingo.

AI assessment note: “a few things were I would say maybe headwinds to starting to monetize”

Q Why? Because everyone says that on the show.

A They love, they love saying that. First of all, the CEO is the CEO of the product, not the product manager. The product manager has to understand what the user will respond to when you kind of give them what they're going to see in the app. So they have to understand the user design really well. They also have to be able to stretch into business strategy, et cetera. That said, this is why consumer products live and die in the pixels and attention to detail matters. If a button that's like the dark shade of green versus the light shade of green will make a difference in user behavior, a product manager or a product designer for that record, like people working on the product should understand what that means. And I think this is also how we build product teams at Duolingo where everyone should understand pixel perfect design and why certain Specifically what the user is going to see, how that's going to change user behavior.

AI assessment note: “First of all, the CEO is the CEO of the product, not the product manager.”

Q Do you lose the design process? Does Figma become obsolete in the world of wrap, let's say?

A This is a really good question. I don't think it becomes obsolete, but the way we use these tools and maybe it will be Figma, maybe it'll be a different tool is going to change significantly. I think we'll go from basically text of an idea to a prototype of an idea rather than everything in between. Maybe Figma is the tool that we use to do that. Maybe it's something else. But then now there's like three, four steps to get there, or actually even more. Like there's basically text to design, and then design to code is kind of how things go. Actually, there's text to spec, spec to design, design to code. This is gonna all merge, is my guess. So this process is changing already.

AI assessment note: “I don't think it becomes obsolete, but the way we use these tools”

Q You mentioned Lily there. I, Obviously spoke to Severin the other day. Wonderful, elegant man. And he told me about like two of the, I think it was PMs who came up with the idea for chess and brought that forward. And they weren't really sure. How do you think about the ideas you choose to pursue versus not choose to pursue?

A This is a great question. You know, I would love to say we have this like super ironed out process, you know, you follow ABC and you have the perfect answer to the question of what to invest in. The reality is it's a lot messier than that, but chess is a really good example of bottoms up innovation in the company where it's not an executive that had a great idea and said, yeah, we should go in this direction. So it was a product manager and a product designer. They didn't have an engineer on the team. They went and prototype. Basically they're like, we think chess will be a good subject for doing it to do. We're like, kind of weird. Doesn't exactly fit, but we're like, yeah, cool idea. They went and prototyped and almost like a full chess lesson experience with no engineers. So like our PM vibe coded basically the whole thing. Our designer created excellent looking designs. We tested it and we're like, this is a pretty good Duolingo lesson. So like they kept iterating actually like on this vibe coded prototype. And then until a point where we're like, this is very much headed in the right direction. We really enjoy using this prototype. And then We staffed a team with engineers, and now it launched this week. So in a span of eight months, it went from like, I don't know, this is kind of an interesting idea to like a full launch of a new Duolingo course.

AI assessment note: “The reality is it's a lot messier than that, but chess is a really good example”

Q How do you think about the right time to hit them with a paywall? You want to build the trust, the love, the loyalty without being like, okay, level one paywall.

A Exactly. So I think it's twofold. One of them is honestly like hitting with a paywall is kind of the wrong approach in the sense that I think you want people to discover the paywall by experiencing features. That is the best way to honestly like convince someone to buy something. Meaning let's say I would like to convince our users to buy Duolingo Max and it's It's key feature is a video call. The best way to do it is have them try video call actually. So a lot of our ways to do that is maybe show them a video of how video call works and then say, if you're interested, here is Duolingo Max or, uh, have them go into the experience, have a call themselves. And then at the end say, this is part of Duolingo Max. That is kind of the best way when it's like contextual and users discover that there is something to be paid to like experience. I think just like having people showing kind of what I would call like generic ads works way less better. And it looks spammy in the app. So we try to do everything contextual. For example, if we show you an ad, we will say right after there's an ad free version if you'd like to upgrade. And that is in a way like you've discovered an ad. Okay. Maybe you didn't love it. Okay. But you can also find an ad free upgrade.

AI assessment note: “you want people to discover the paywall by experiencing features.”

Q You mentioned Lily there. I, Obviously spoke to Severin the other day. Wonderful, elegant man. And he told me about like two of the, I think it was PMs who came up with the idea for chess and brought that forward. And they weren't really sure. How do you think about the ideas you choose to pursue versus not choose to pursue?

A This is a great question. You know, I would love to say we have this like super ironed out process, you know, you follow ABC and you have the perfect answer to the question of what to invest in. The reality is it's a lot messier than that, but chess is a really good example of bottoms up innovation in the company where it's not an executive that had a great idea and said, yeah, we should go in this direction. So it was a product manager and a product designer. They didn't have an engineer on the team. They went and prototype. Basically they're like, we think chess will be a good subject for doing it to do. We're like, kind of weird. Doesn't exactly fit, but we're like, yeah, cool idea. They went and prototyped and almost like a full chess lesson experience with no engineers. So like our PM vibe coded basically the whole thing. Our designer created excellent looking designs. We tested it and we're like, this is a pretty good Duolingo lesson. So like they kept iterating actually like on this vibe coded prototype. And then until a point where we're like, this is very much headed in the right direction. We really enjoy using this prototype. And then We staffed a team with engineers, and now it launched this week. So in a span of eight months, it went from like, I don't know, this is kind of an interesting idea to like a full launch of a new Duolingo course.

AI assessment note: “The reality is it's a lot messier than that, but chess is a really good example”

Q Do you lose the design process? Does Figma become obsolete in the world of wrap, let's say?

A This is a really good question. I don't think it becomes obsolete, but the way we use these tools and maybe it will be Figma, maybe it'll be a different tool is going to change significantly. I think we'll go from basically text of an idea to a prototype of an idea rather than everything in between. Maybe Figma is the tool that we use to do that. Maybe it's something else. But then now there's like three, four steps to get there, or actually even more. Like there's basically text to design, and then design to code is kind of how things go. Actually, there's text to spec, spec to design, design to code. This is gonna all merge, is my guess. So this process is changing already.

AI assessment note: “I don't think it becomes obsolete, but the way we use these tools”

Q When they're settled users, do those cohorts in terms of usage look quite like Candy Crush users, like the very, very settled gaming users?

A I think so. I'm not knowing Candy Crush's metrics myself at all, but what we intend to see is if you've reached a 30 day streak, so you've built up a 30 day streak, your retention is very close to a hundred percent. It's not a hundred percent, but it's very high because you've built a habit. Like, it's not crazy because you've done it for 30 days without breaking your streak. You have built a habit, so you're very likely to continue with that habit. But if you look at retention of users who are streak less than seven days, It's way lower, because they actually haven't built a habit. Or maybe they're unsure of doing it, or they're just playing with it. One way or another, they haven't built a habit.

AI assessment note: “I think so. I'm not knowing Candy Crush's metrics myself at all”

Q When they're settled users, do those cohorts in terms of usage look quite like Candy Crush users, like the very, very settled gaming users?

A I think so. I'm not knowing Candy Crush's metrics myself at all, but what we intend to see is if you've reached a 30 day streak, so you've built up a 30 day streak, your retention is very close to a hundred percent. It's not a hundred percent, but it's very high because you've built a habit. Like, it's not crazy because you've done it for 30 days without breaking your streak. You have built a habit, so you're very likely to continue with that habit. But if you look at retention of users who are streak less than seven days, It's way lower, because they actually haven't built a habit. Or maybe they're unsure of doing it, or they're just playing with it. One way or another, they haven't built a habit.

AI assessment note: “I think so. I'm not knowing Candy Crush's metrics myself at all”