Dynamic Paywalls That Drove Millions in New Revenue – Shawn Gong, Tinder

0:00I think maybe the simple way you can do is design three products or tiers for the customers. Why? Because that's a very common or easy way for us to pick from. Some of one want to buy the most expensive one. Some of just want to buy the cheapest one and then some of one don't know how to make decision and buy the middle one. So that's usually the simplest way for you to design your product is because actually change how decision are made. Let's say David if we only offer one product. So your decision is should I buy or not. But if a day will show you three products now you are

0:37thinking which one should I get? So you bypass the yes or no and then more likely to convert. Hello, I'm your host David Bernard. Today's conversation is shorter than usual and will be featured in Revenue Cat's state of subscription apps report. Each episode in this series will explore one crucial topic and share actionable insights from top subscription app operators. With me today, Sha Gong, a product leader building monetization and growth engines for consumer unicorns, including Tinder and Grinder. On the podcast, I talk with Shawn about how Tinder's machine learning powered payw walls drove millions in new revenue. The art of selling features all a cart

1:20without killing subscription revenue. and why Tinder Select flopped despite users saying they'd pay for it. Hey Sean, thanks so much for joining me on the podcast today. Thank you so much, David, for having me. It's such a pleasure. I always love your newsletter, podcast, and all the content. So, it's it's what an honor to

1:41be here with you. Oh, thanks so much, man. Well, I so I've had several ex Tinder employees on the podcast, but super excited to get you on because I haven't gotten to talk to somebody who's been in the trenches the last few years and Tinder's done a lot of cool stuff on the [snorts] monetization front. So, I wanted to dive into a project I know you worked a lot on and that's AIdriven pricing. So, what was the problem? What were you trying to

2:06solve and and how did things go? Yeah, I'm very excited. So let me start with the core problem because that's really you know shaped everything we build at Tinder. Um so the user problem is decision overload simplified to put that way. So as you might know Tinder has a lot of purchase options from multiple subscription tiers and then under each tier there is a multiple plans like weekly monthly and then we have a lot of allocart products. So it's great right because we have so many product for user choose from. However we talk to users and we notice two things at least. One is some user they bought

2:47the platinum the highest subscription tier we offer to everyone simply because that's the most expensive one. So their belief is that well I'm willing to pay the most and then I'm gonna get the best so I'm gonna buy platinum which is great. However, we noticed some of the users who bought platinum, they only used the features under our second tier, gold. So, they realized, oh, actually, you um bought something you didn't even fully use. You could have just bought our gold subscription tier to save you money. Or maybe another issue can be like we can do a better job to educate users. Hey, there is other feature from

3:25Platinum. You should take advantage of them. You haven't used them yet. Another thing we noticed is some user they didn't buy anything simply because they were overwhelmed. It's like oh my god so many options I don't know what to pick. And then also reality we learn is maybe David you have your own experience you know when you look at particularly you revenue c specialist in pay wall right so one common mistake a lot of company make is they describe all the features or benefits the benefits are better than feature yes however do you think user going to read through them make a decision no that's what we thought user

4:01going to do okay let's compare tinder plus tinder globe tinder platinum and then compare the other car and then make decision. No user make decision within a second. So it's our job to help them to make decision better. So that's from the basically user problem is like decision overload and then so it's overwhelming. And then from the business problem is of course that leads to relatively lower conversion right because for some user who wanted to buy but they were overwhelmed they decide not to buy and then that a lower conversion. So that's a classic case of misalignment um incentives. Um so we wanted to maximize revenue but our UI experience made it

4:46harder for user to make a confident decision. Yeah. And I mean Tinder famously I mean this is why I have you on and so many people in industry have talked about Tinder because it's led the way in these hybrid monetization but having three different levels of plan and and then you you know you'll test it out a fourth level of plan. We'll talk about that toward the end hopefully if we have time. But it does get super confusing and then you have the super the booths and the other inapp purchases and things like that. So then what did you do to actually to solve that and to

5:15to make it more accessible to folks and getting people to the right plan? So um that was a challenging but luckily we had a brilliant uh machine learning team. So when I talked to the team members and then they told me hey we can try to build ML models and then use that to predict users willing to pay and then we can surface the best product they might most likely to buy. So that's our solution because based on the insight we talk about is like people customers don't need a lot of option they need the right one right think about like Netflix you know how many users like oh my god I

5:53don't know what to watch there's so many choices they end up spending like hours scrolling and then not making watch anything that's why Netflix have something like top picks for you right based on your previous behavior and your ratings they predict what you're most likely to to watch so solve that problem so very similar So this is a huge moment for us I think you know particularly even for Tinder for the industry is we shift from uh stale pricing to like dynamic pricing so we don't show the same pay wall same product to all the users um so you know think about let's say David you are willing to buy

6:32platinum why we should we show you plus right I mean that's a problem not just from business perspective we won't be able to maximize the revenue And then we can use that to reinvest for to make product better for you. Another problem is you don't get to enjoy all the best benefits we can offer right simply because you chose the plus. Um so that's what the solution come within. We use a machine learning we train the model and then we and eventually release a model for that to predict and serve the best skew for customers. How do you test that? Like what was the what was the AB

7:08test? It was just the standard kind of deterministic normal everybody gets this payw wall at this part of the flow and then everybody gets that pay wall at that part of the flow. And did did you just do an AB test to prove that the ML decision model was was better than what you had before? Yeah, spot on. Yes. Um but obviously to reduce the risk um we couldn't just like

7:36start it from all the pay walls, right? because we have a lot of different payw wall and then the same times that's very expensive to uh and it takes a lot of time and then effort to train the model to test the model. So we start with something small. So we test with couple features um and then just you know start with not all the scope. So we can see hey based on this machine model to um historically you know like a version a control will be same you know like as we used to show the payw wall and the products to the new one this is dynamic

8:12we will change based on the basically when they show a user let's say for David and then our payw wall will ask the ML model hey which payroll should we show and then the pay will decide okay which product we should recommend to David and then That's how we test this treatment and then we'll be able to measure the conversion and then the

8:32total revenue and how how did it do? That's great. Yeah. So I Yeah. So based on our prediction, I think it's definitely multi-million dollar annual increase uh for Tinder. Also want to clarify, you know, for that is like it's not just like maximum revenue. It's like a evil business plan. know it's like the money we're going to use we can reinvest to the product right so we can improve our user experience and then build more like benefits for customers that's why uh when I talk to a lot of startup founders I always advise them hey don't wait to think about your monetization strategy you know you don't feel bad to

9:13charge your users they wanted to pay you and they deserve it because you're going to provide better experience for them not mentioning to compete with your uh competitors and you'll be able to move faster and sooner. What about counter metrics? Were there anything you were watching to make sure things didn't go south in retention or user experience? You know, that people weren't happy with with the plans that

9:36they were uh presented. Tinder has a really good process in place and then we are required to also measure our contract metrics. This is great because you know this you cannot only focus on on one area for example for this case we cannot just only focus on the first time conversion and then only the revenue amount right so we have to measure hey let's say David based on our model maybe prior to that he would buy a plus now he bought platinum so we want to check hey is David going to come back is David going to buy platinum again it's going to cancel so we

10:14definitely measure this long- term term success metrics to make sure this model really served correctly because that's nothing we wanted to continue to work on like a most product features right it's not like launch and down that's it like we have to optimize iterate so we wanted to learn hey we actually have different models we built a different ones we want to compare so we want to see how they perform between those model same time how they perform compared to the the pay wall without the models. So, we can make sure we focus on long-term success. So, the takeaway for this example is the real unlock wasn't just better pricing,

10:56it was a better decision design, right? Helping users to choose a product they truly fit them. Um so even for your product you know any founders or startups out there even you don't have a ML team yet um I think maybe the simple way you can do is um design three products or tiers for the customers why because that's a very common or easy way for user to pick from right someone want to buy the most expensive one some of just want to buy the cheapest one and then some of one don't know how to make decision they buy the middle So that's usually the simplest way for you to

11:36design your product here is because actually change how decision are made. Let's say David if we only offer one product so your decision is should I buy or not but if a day will show you three products now you are thinking which one should I get so you bypass the yes or no and it's more likely to convert. Yeah that's fantastic advice. Um I I do wish we all had ML teams. Maybe that's something at Revenue Cat we should be working on to help folks build those kind of sophisticated things into their own apps. But anyways, I did want to move on to this idea of monetization

12:11unbundling is that you know not everybody wants the full subscription. Not everybody is going to fit into the gold the platinum. So yeah, how do you think about unbundling that? We actually you probably know we had some um bundling products already car such as like boost super likes I think a couple thing you want to think about like besides one issue you've already highlight is not everyone want to have a subscription right subscription is great I mean don't get me wrong because you get a package it's much easier than you have to decide which features you wanted to pick from and then to buy that way

12:49it's it's painful right it's easy to buy a whole package package and then repeat it uh renew automatically. would have to think about twice but there is some use cases think about like boost I think it's perfect allocar product why because you don't it depends on when you want to use it right when you don't receive enough likes so you want to boost yourself when let's say you swipe during the peak hours so you wanted to use that so that's makes sense super likes too right because it's like well I don't know how many super likes I'm going to send it depends on who I see in the app

13:23right and another use case we decide to try is the travel mode also aka passport mode. It's allowed any users to see anyone globally. So it's so cool you know like let's say you can go to like Paris, go to Spain, whatever you want to go based on your needs can be basically you want to travel there or you want to just meet people there and the thing about that feature is like you may not everyone wants it right maybe because it's very special case for needs for certain people in that case that feature is a great candidate as like a unbundled feature stand alone. So that's how we

13:59think about what features make sense to be unbundled and designed for our customers. How does a passport mode work? Is it a a one-time purchase for a limited time? Is it an add-on subscription or how does it work exactly? Yeah. So, historically, this feature is part of a subscription. So, you have to get a subscription to use it. And then, but like we talked about, the challenge for that is for some users, they only want to use it when they plan to travel or to meet people outside of their home city. And then they might not want to buy a subscription, right? because they just don't want to have that commitment or

14:35they don't want to they don't feel like they need to use other features. So that's a perfect candidate for them. So we test out the features. So that helped us able to capture nonsubscribers for the users who just want to buy all the car for this passport feature and then they can use it for like one day, three day a week based on the offers we provided and then they can enjoy that benefits to meet people anywhere in the world. So, how have things been going with the the passport feature and it

15:04being unbundled from the subscription? Have have you seen uptake? Yes. Um, actually I I think it's it's so fun to talk about this because it was not launch and then success story. That's it. Not that simple. We learned a few things based on our test. First thing is when we launched this um standalone allocar passport feature and then the

15:27conversion went crazy. It was great. However, we noticed like, oh, that hurts our plus subscription. [laughter] Uh, because some user decided only buy this feature, not to buy subscription. But that can be a couple things. So, one is, oh, maybe we price too low, it's too affordable. And then so it's like no brainer to user use it. The second thing we did is like, oh, we should increase the price for other card and to see how that would change the conversion and then subscription cannibalization. So we definitely noticed that the sysun can cannibization reduced as we expected and then the conversion reduced a little bit but total revenue actually went up. So

16:06that was very positive and then we continue to test because we wanted to minimize the subscription kind of positions. So and then we increase the price again and then also have upsell on the pay wall. Basically, we show, hey, David, you can buy passport um standalone alloc card, but also you can buy plus subscription. And then we use a very interesting like a kind of psychology thing. Basically, for our 7day passport feature allocart, the price is the same as 7-day plus subscription. So, that case, David, you think like, oh, duh. Then the 7-day plus subscription is a better deal, right? I can get it more than this feature. So

16:48that also helped of course you know means like reduce the cateization also increase the conversion or revenue but it's still not good enough. So we decide hey let's use our perfect ML model again. So we will show users subscription first. If you don't want to buy subscription we show you all our car. So we give you a second chance. I love hearing these kind of stories because I just think Tinder's been around so long and and so many of us in the industry look up to it as a pioneer in monetization in apps and I just think so many apps are leaving money on the table by not experimenting with these

17:26kind of things and not trying all the different you know three tiers and you know and and like we said it can lead to confusion. It's not the right thing for every app, but there's there's just so much opportunity to uh meet users where they are and and charge them. And then speaking of which, Tinder Select was a really interesting experiment. So, I'd love to hear from you how that went. You know, it just sounds fantastic and we've we've talked about on the podcast many times before, you know, add a super premium tier. You know, we talked about it earlier like some people will just buy the most expensive thing. So,

18:00sounded like a really interesting idea. How did how did things go with Tinder Select? Yeah, I think Tinder select a very interesting case study in terms like the product market alignment and the behavior science. So let's start with the hypothesis. you know basically the opportunity Tinder was exploring you know why Tinder created uh Tinder select because Tinder has a massive global user base and then we also have some whales right you know basically the user who are willing to spend a lot of money they buy a lot of highest subscription and then all the cards they spend a lot of money with us um to maximize the outcome

18:34so we definitely saw hey there is a willingness to pay out there let's try if we have a even higher end tier and then provide even better service and product would the user buy them not um so that's basically the hypothesis we had and then so that would be different segment we're targeting obviously right and then same time for the I think the challenge are two things one is the identity fit and the brand positioning right you know because for any product it's not every company can just offer a high-end it's really rooted into like what is your and is and then how customer thinks like a good fit or not

19:16right I think for Tinder because we catering for the massive population and then in that case it's um little bit difficult to for some user feel like oh you know it's good fit for me to use it think about that David if we tell you hey you're going to pay $499 a month for subscription and then for on Tinder and then when you come to Tinder you see people you might be able to see even just use a platinum subscription. Now, you might think like, okay, I don't feel that special here. Maybe you give me some special treatment and benefits, but I don't feel like the environment people

19:53here that that special, right? That's just difficult. That's very common. You know, think about like a club too, right? It's like exclusive one, very expensive one versus just like a regular one, but you have expensive table. So, think about that scenario. I think that's why make it challenging. It's like a luxury offering, but people might not feel the benefits meets their

20:14expectations. What's the future of Tinder Select? Is it is it going away? Is it going to evolve? Yeah, I think this we're just not going to expand much further and then just uh gradually scale down for this one. It's not like a fail is basically like we learned what we initially wanted to and then we realized it is not worth the continue investment and the mitness for us to have this feature. It's not the best fit for us. We have better bets to continue. Well, I think it's super cool that you all tried and uh and a great lesson as always that people don't always do what they say in a user survey

20:53or a user interview. So always a great uh lesson to remember but you should try and that that's the great thing and then you did and you learned. Is there anything you wanted to share with the audience as as we do wrap up? I think the most important thing I want the audience to realize or understand is don't treat your users as they are logical human beings. So what I mean by that is you cannot think you cannot be like oh let me design this product and the user going to read everything on the screen they gonna to make decision based information I share with them and then

21:32make a purchase decision whatever they want you wanted them to do right no reality they don't do that way um they don't follow your user experience they don't just read everything to make decision so they are like us you know we make emotional decisions I think that's very important. So that's why we have to really observe how they behave and talk to them and to really

21:55truly unlock the emotion behind that. Otherwise you we're going to make wrong decisions. Where can people find you or are there any jobs you wanted to shout out at Tinder? Oh yeah. I think Tinder definitely has a still growing has a lot of fun opportunities. So encourage people to check out. It's one of the best places I've ever been because it's really fun culture, very progressive. and then really care about people's life. When I go to the cafeteria, every time I get emotional because I see the the wall, we show all the user who wrote letter to us, show their wedding pictures. Oh, you know, like our work, you know, made a

22:30difference, you know, made people having a new family together. How how lovely that. And then same times also if you wanted to grow your product and then without hurting your retention and then uh I can help you to unlock user emotions and design defensible growth loops and then you can find me on LinkedIn and then mental cruise. Awesome. Thank you so much for joining

22:54me. This was a really fun conversation. Yeah, thank you so much Davis. It's always my great great pleasure talking to you. Thanks so much for listening. If you have a minute, please leave a review in your favorite podcast player. You can also stop by chat.subclub.com to join our private community. [music]