We haven't got a good enough prototype to even validate if it's a good thing. David Barnard (00:15:45): Yeah. And then what do people spend most of their time doing anyway? They're on Instagram looking at pictures and videos and things like that. Jacob Eiting (00:15:51): We don't need that. You know what I mean? There's not much more bandwidth to be squeezed out of a next form factor, right? The phone, yeah, it's a very high bandwidth. It's very good form factor. It's well engineered. David Barnard (00:16:03): And if it does move to glasses, it will just be like an evolution of the phone in a way. Jacob Eiting (00:16:09): Look, you could have called in 2007 that you wouldn't have your Macs anymore, but I think pretty much everybody ... With the exception of the casual computer user who is just browsing email, anybody who's serious about compute in any way still uses a computer, right? And I think even most people still use a computer. I don't know what the penetration numbers are actually. I think I'm 80% bullish, but I want to leave myself ... Stranger things have happened, right? Stranger things have happened. Who knows? Maybe when the metaglasses, the AR glasses get really good, we have a different opinion, but I don't know. There's not a strong track record right now of any of these devices going completely viral. But you would say that too in 2005 about smartphones, right? David Barnard (00:16:49): On Twitter and kind of in the forefront of all this with us tech nerds, it does seem like there's a lot of people who think it's going to happen a lot faster than it really is. Jacob Eiting (00:17:00): Is it the singularity? David Barnard (00:17:01): No, more just that like, oh, the phone's going to be absolute Jacob Eiting (00:17:04): Apps are going to be- We'll call that the app developer singularity for all intents and purposes. David Barnard (00:17:10): Apps are dead is like a saying for the past 15 years. Yeah. Jacob Eiting (00:17:15): Cursor's dead too. Everybody thinks things are dead. David Barnard (00:17:18): It's all going to happen much more slowly than you would think reading Twitter. I mean, even I get on the hype train of like, "Oh, robots are going to be doing my laundry and dishes." It's like, yeah, eventually. Jacob Eiting (00:17:32): 20 years. Yeah. I'm not going to make any bets on the state of subscription apps in 20, whatever that is, 56. No, you're 46, 2046. David Barnard (00:17:40): But we've got some time to keep building these little apps that help people. Yeah, Jacob Eiting (00:17:45): Yeah, yeah. And I mean, look, I think developers are going to go where the economics are. People came to the app store because it was the place to build where they could make money. That's not going to shift completely overnight. And I still, I called it to somebody yesterday, the wet layer. I think there's going to always be a wet layer of software between the human and the human. I just think there will be because there's always going to be that last 10% that a bot can't just like one shot that OpenAI doesn't quite understand natively in the model. There are just certain things where like even today that we could automate, but we don't. You know what I mean? Because people want people in the loop, right? Like sales, like could you automate sales completely? Yeah, people try. It's like called self-serve, right? But there's still this like niche, I mean you call it niche, it's like you're selling big ticket items. (00:18:35): People want a wet layer. They want like a human in the loop, right? And it's not necessarily, doesn't even make it better, right? It just makes you feel. It's a feelings thing. I don't know what that looks like in software at scale, but it's plausible to think that there's at least a case. But I don't know, David, this could be a podcast by horseshoe salesmen in 1910. Do you know what I mean? Like convinced, "Oh yeah, it's going to change, but it's not going to be that fast. These things are loud and noisy and they break down all the time." They talk about cars. So I always want to leave the possibility that we're just like out of step. And I think if anybody of our listeners are worried about it, could be worried about it. I mean, you're playing in a market. Anytime you're in a market, you need to be smart and you need to be eyes up and you need to be ... If you want to win and compete, you have to be thinking a little bit ahead. (00:19:20): So you should be trying these tools. You should be looking at what's going on in the market. You should be, at the same time, not losing the fundamentals, which is like helping customers achieve their goals and talking to users and like all of these things. I think that's maybe one anti-pattern I've seen is like, and maybe the open call rage has been a thing where it's like, "Oh, finally an AI is going to run my company, get me to 10K MRR, make no mistakes." And that's cool. There's definitely leverage and open call and things like this. But I think the problem with free leverage is that everybody gets it. So it's like very quickly, the alpha, the advantage it gives you over the market is fleeting, right? So as long as you have some information that other people don't, you know something that other people don't. (00:20:05): That's how you get lasting alpha and lasting competitiveness in the market. So anyway, well, it's been a great podcast, David. Thank you. We're done here. David Barnard (00:20:13): We haven't Jacob Eiting (00:20:13): Even David Barnard (00:20:13): Talked numbers. Jacob Eiting (00:20:14): All right. The boring part now. We numbers. We don't need to talk numbers. No. David Barnard (00:20:19): I know most of you who read the report are just going to skip over the methodology section, but the methodology section is bigger and better than ever before. And we actually explain in a lot more detail than we have in the past a lot of what went on behind the scenes to generate the numbers, what they mean and stuff like that. And so to better understand the report, spend a few minutes, read the methodology. I'll give a few highlights. In preparing the report, we didn't just look at all data. We did filter by apps that have active subscription revenue and meet minimum thresholds of installer revenue. So we did actually lob off the kind of ... I don't even know ... I should have asked this. I don't know if it's like 10% or 25%, but this is not data from the app that's only gotten 10 installs and one subscriber or something like that. Jacob Eiting (00:21:10): We do some data preparation certainly that we think is appropriate to getting good signal. David Barnard (00:21:16): And then for RevenueCat customers to understand, it is all fully anonymized and aggregated, so you're not going to see any individual's data in there. I've begged for years to figure out some way to include scatter plots because they're an easier way to really understand distributions, but that would technically show individual user data. So we don't do scatter plots. You Jacob Eiting (00:21:35): Could always do a distribution and then just make up a fake scatterplot. Sample Monte Carlo style for 2027, we'll do that. David Barnard (00:21:44): And then one interesting note this year, and definitely worth kind of reading the description in more detail, but we separated AI apps from non-AI apps to just kind of get a sense for what's going on in the market. And Jacob Eiting (00:21:57): We've emailed all of the investors of non-AI apps. So it's time to get with it, get with the program. David Barnard (00:22:03): And so just as an overview, whether an app uses AI or ML models for its primary value is kind of how we're trying to ... And it is fairly subjective. This is not an objective measure. There's not like a checkbox people do in the app store of like, "This is an AI app." Jacob Eiting (00:22:19): You know when you see it, David Barnard (00:22:20): That Jacob Eiting (00:22:21): Sparkles, it costs more. David Barnard (00:22:24): So that part is not perfect, but we did our best and it is interesting to kind of make those comparisons. And some of the comparisons, and we'll talk about it, is that AI apps are commanding a much higher average revenue per user and things like that. So it is interesting to break it down by those. One other thing to note, this year, our production team is going to be inserting slides in the YouTube video. So if you've gotten this far along on the audio only podcast and want to jump to YouTube, we will have slides show up in YouTube as we talk through this data. So it might be helpful for those of you, but we'll try and explain it and talk about it in favor terms. You don't have to look at the charts. First chart, apps are booming. We kind of already talked about this, but when you look at the chart, it's freaking wild. (00:23:15): In 2022, in January of 2022, 2,000 new subscription apps were launched per month. In January 2026, 14,700, over 14,700- Jacob Eiting (00:23:29): So a 7X? David Barnard (00:23:31): 7X increase in the number of Jacob Eiting (00:23:32): Subscription apps in a month. It's a lot. I can concur. We are seeing a similar thing. Actually, and most of that growth is concentrated in the last six months, I think. David Barnard (00:23:44): One thing to note real quick too, this is data. These two slides we'll talk about at the start are actually from AppFigure's data. Yeah, it's not our data. This is the entire market. This is not just revenue cap, although we're shipping in a large portion of that. Very proportionate Jacob Eiting (00:23:58): To that growth. I mean, we certainly have tracked that curve, but yeah, I think, I mean, this goes back to what we were just talking about, but we're in unprecedented times, as they say. But yeah, again, it's just like the natural ... What happens when you drop the cost of production massively, you get an oversupply ... I mean, they call it an oversupply. You get a massive supply boost on apps, which is great. I think we've seen Apple trying to catch up. App Review's been bogged down. We've been obviously trying to handle all the interest and demand. Interestingly, this is B2B insider stuff, but these new customers are easier than old ... You know why? They talk to the bot. The bot does everything. They don't even ... Again, API, I don't even know we're a company really. They just set the account up there. (00:24:42): The wet layer just does the OAuth thing and we're done. But we've not seen an increase in our ticket volume, which has been interesting. And also we've gotten just better. The product gets better, so tickets go down, but it's been really interesting. Well, we've David Barnard (00:24:54): Done a lot of work building their RevenueCat MCP and we've put a lot of work into making it really easy for Jacob Eiting (00:25:00): Agents David Barnard (00:25:00): To do the implementation. Jacob Eiting (00:25:03): In terms of watching ... It's not really relevant for this crowd, but it's a good anecdote on why the charts that and ratios you got used to in the old world don't really apply because things change. You wouldn't expect ... I think a thousand more ... 7X more users, seven times more tickets. Nope, we've got seven times more users per month and same number of tickets. It's like, "What?" And then you're like, "Oh, it's because they're already talking to an AI." So they just keep talking to the AI and can solve 90% of those problems for them, which is great. David Barnard (00:25:33): One thing I was surprised in these, I figured numbers is that iOS now accounts for 77% of an all new subscription app launches, and that's up from 67% in 2023. I would've thought that if you're vibe coding an app, why not just use React Native and launch it on both platforms? But apparently a lot of folks are just either going native or still just not bothering to launch on Amtrak. Jacob Eiting (00:26:00): I mean, one thesis, I'm just riffing, but could be the app store Schlep, like getting all your accounts approved, getting all your stuff in place, your bank accounts, da, da, da, da, has become a larger proportion of the painful part of building an app. So only doing that on one store reduces ... You don't have to also go do that on play, right? So maybe that's part of it as a ratio of total work, but also it's not a huge shift, like 66 to 70 or whatever it was. David Barnard (00:26:26): Yeah. Gosh, I didn't even think about that. But no, that's a really good point. If you're vibe coding an app over a weekend and the agent's doing most of the work, if two hours of figuring out how to create a Google Play account and entering your social or setting up a company and getting your Dunes ... You got hung up on getting a Dunns Jacob Eiting (00:26:46): Number, I know. Yeah, I'm still hung up. I'm two months into getting a new app store account set up. Never going to happen. I'm done. I'm retired. David Barnard (00:26:55): It's funny that just setting up an account is one of the harder things in building an app. Jacob Eiting (00:27:00): Yeah, the bottlenecks have shifted. So I don't know. It's interesting. David Barnard (00:27:05): Another chart we included from app figures is the share of revenue in 2025 from cohort of when the apps were built. Am I explaining that while you want to take- Jacob Eiting (00:27:17): How old are the apps? David Barnard (00:27:19): How old are the apps that are making money? Jacob Eiting (00:27:21): What changed? David Barnard (00:27:23): Older apps are making most of the money. I mean, and it's not a change. It's just like a state of what's going on today. 69% of revenue in the app stores is coming from apps that were released before 2020. Jacob Eiting (00:27:36): I almost commented on the doc to be like, most of history is in the past, news at 11, right? It's like, of course. But winners win, they compound. And I think we'll always ... I don't know if it's shifted precipitously. I think you could imagine a world where more revenue is coming from newer apps would be a story in terms of a higher dynamicism in the app store, just more to mult. But it's just like, come on, you're not going to ... Just the compounding advantage that Strava and KLAS Dojo is a good example maybe. And I was also thinking Duolingo, all these ... And then of course, all of the media companies- And Netflix and Paramount. Those are going to just continue. They're not going to shrink. You know what I mean? Most likely. They might trade a little bit back and forth. I would love, so we should go back and do so. (00:28:24): We could reproduce a Sosa 2020 because I think you'd probably see the same story and be like, "Most of the apps are proved before 2015." It's like, yeah, of course. David Barnard (00:28:33): Yeah, it just takes time to mature. The interesting thing on this chart though is that you do see 2022 and 2023. They kind of bucketed it weird, but 2020 to 2021 is 9% of the revenue, 2022 to 2023 is 14% of the revenue. So there was a bump in 2023, and I imagine that's actually mostly ChatGPT, OpenAI. Jacob Eiting (00:28:53): Yeah. I mean, I don't know. It's app figures numbers, right? So 2023, it wasn't just ChatGPT. It's like all of these, in the next 10, a lot of these foundation models launched their products in 2023, which yeah, that's probably what you're seeing there, which is interesting.That is interesting. When you see a displacement like that where the older cohort's actually producing a smaller share, that kind of really illustrates the shift that we've gone through. David Barnard (00:29:20): And this is going to be really interesting to follow up on in the coming years because this is the chart that's going to show you disruption. Are these newer cohorts of apps disrupting older apps and taking bigger shares, growing faster, capturing wallet? Jacob Eiting (00:29:37): A supply demand question implies the demand is limited, right? Which if the demand's growing fast enough, everybody's growing and just it's different rates. Yeah, it's hard to visualize. David Barnard (00:29:45): So an interesting one to look at now, and I think it'll be even more fascinating in the next couple of years, but speaking of growing faster, that's the next thing I wanted to talk about was we have a chart in here talking about year over year growth in MR RR. And this is fascinating. And it's just a perfect illustration of power laws is that the top 10% of apps grew 306% and the bottom actually lost money and the median app only grew 5.3%. Jacob Eiting (00:30:19): What was the GDP growth of the last year? (00:30:22): Top 5.3%. Isn't that funny? These things level out, but that's a median. It's not an averaging of all those things. Actually, probably if you average the average of the app source much higher because those outliers just contribute that top 5%, that top 2% of MRR growth last year probably drives a huge chunk of the total MRR average movement and that won't affect median as much. But yeah, I mean, that's winners and losers. The sorting machine, that's what Stripe called it a bunch in their annual report this year is the sorting machines. I don't know who they didn't come up with that, but these markets are just sorting machines and they're going to sort the fittest to the top and the least fit to the bottom. And yeah, I mean, we've seen, again, I'll draw B2B. We've seen just insane meteoric growth rates at the very top tippy top of companies. (00:31:10): Cursor just announced yesterday that added a billion in revenue in a month or in a quarter and that company's only two years old or something like this. Totally unprecedented. We're seeing that too in these AI apps. If you were the app that has the model that's the best, you can just get numbers that never were possible before because just that level of differentiation was never possible before. You could be the best whatever. I use my friend Flighty, the best flight tracking app ever. I don't know if they ever had that level, you know what I mean? And stuff. It's just like you have this ability that if you've done the CapEx capital expenditures to build one of these models, when you go to monetize it, it's truly better. I mean, that's why the CapEx spending is so crazy. It's because it's like the returns are there. (00:31:56): I wouldn't anybody, I think unless Sam's listening or Dario's listening, or I don't think anybody should feel too bad if they're not in that top, top, top category. There's a lot of ... That means half the apps are above 5.3, but you're beating that GDP. You should be kind of happy about that in some way. Now, if you lost 33% last year, it's like, ooh, maybe rethink your strategy. David Barnard (00:32:19): Rich, get richer, poor, get poor. I Jacob Eiting (00:32:21): Mean, that's David Barnard (00:32:22): Just the way things Jacob Eiting (00:32:22): Go. It's probably people's apps they've abandoned. You have to think about that too. It's like just because. Oh, I'm in David Barnard (00:32:28): That stat. Jacob Eiting (00:32:28): Yeah. If you have an app in this dataset, you've written it off for whatever reason. You might just be moved on. It might just not be the best ROI anymore. So that's always in the dataset too. So these are not all strivers. These are not all apps that are trying necessarily. David Barnard (00:32:41): I had a big launch in 2024 that did really well and then we did not follow that up with marketing. Jacob Eiting (00:32:48): Yeah. I mean, basically treading water implies effort. It implies you're continuing to adapt and whatever the new marketing thing is and whatever the new features thing is. David Barnard (00:32:57): But it is interesting that the top 25% of apps grew 80% year over year. And it's a bigger market and there's more money. And so there are more apps making more money than ever. And as concentrated as it is in those top apps growing 300% year over year, 25% of apps growing 80% year over year, there's a lot of growth happening and a lot of apps making a lot of money. Jacob Eiting (00:33:21): Yeah. I mean, just think of the return on capital, right? Even if you're spending most of that, I don't know how the economics of the CalAI deal went. I though David, you're trying to get them on the podcast. Maybe we can get some more details. But if you think they probably, I don't know, or maybe margin break even the whole time, right? Just reinvesting in growth, but then you have a big liquidity event at the end, that's a way to monetize that all of a sudden and stuff. So it's like if you can in any way drive 80% growth, even if it's breakeven, that's probably a good bet. If you think you have something with some amount of durability, which like a good fitness app might be, right? David Barnard (00:33:55): And with subscriptions, it's that compounding. If you have reasonable retention, the million dollars of ad spend this year, let's say you have the median retention, which is near 30%. We'll just say 30, I think it's 27. But if you have median annual retention and you spend a million dollars here in March of 2026, then March of 2027, that's $300,000 in free cash flow. And I've talked to a lot of apps- Jacob Eiting (00:34:22): If you have a 1X CAC to LTV or whatever, first year payback period. You can David Barnard (00:34:25): Profitably spend a million Jacob Eiting (00:34:27): Dollars- Or even break even. In the first year. Yeah. David Barnard (00:34:30): Yeah. Just barely break even. That next year, that's just free cash flow. Jacob Eiting (00:34:34): Yeah. And those retention rates get better if the product is sticky. And you also build ... I mean, there's all these compounding things you build. You build a list. Now you have even all your churn customers become people you can market to more or less in perpetuity and all of these things. So I think there's still apps, and this is maybe a meta thing, but apps is a venture target. It's still a very rare asset cloud. I don't think there's a ton of apps that's like venture capital is the right financing, but there's certainly still good financial engineering that you can do to grow these apps with capital very quickly and it can pay off in some cases. David Barnard (00:35:09): It was really interesting talking to Olivia more from Andreessen Horowitz, though she's pretty bullish on consumer. I feel like there's- Jacob Eiting (00:35:17): Yeah. If you can make a hundred bets. David Barnard (00:35:19): Well, yeah. Jacob Eiting (00:35:21): Yeah. It David Barnard (00:35:21): Keeps going in waves. And what she shared and what we're seeing is some of these consumer apps are growing faster and exploding in ways. And if you lump in some of these lovables and replets and other stuff, a lot of that growth is actually consumer growth of people who are tinkering with these and playing with them and using them for those kind of things versus not strictly B2B growth. So yeah, it'll be interesting to see though how venture capital, PE and others treat consumer moving forward as the market grows, as competition gets harder. Jacob Eiting (00:35:56): I think probably right now, at least on the PE side, everybody's going to wait a little bit because I just don't think everybody really knows. There's a lot of B2B software roll up PE that's probably of questionable valuation at the moment. That might just be a January, February 2026 thing. I think so. When growth curves are unpredictable, that's actually when venture should be playing, right? Yeah. But then when we kind of understand these financial assets a little bit better, that's where lower risk portfolio builds like PE and stuff do more. David Barnard (00:36:26): And that's where I do feel like, and again, I talked to Olivia about this in the minisode we did, is that it is a blue ocean now of how do you rethink all of these entrenched consumer apps? Like what's the AI first Calm? And Calm's not necessarily going to be the one to build that. Jacob Eiting (00:36:45): You have to make the choice. If you're that firm, you have to make the choice to reinvent yourself and to probably break some of your precious eggs or whatever in the process. And that historically not always been the case that firms have adapted. David Barnard (00:37:00): And that's where 2026 and 2027 might be the years we see the next ChatGPTs where the launch this year and get to hundreds of millions of revenue in months. Jacob Eiting (00:37:10): Claude's on a run. Their Super Bowl ad paid off. Anthropic's making a real run of consumer already. We thought this was like a one category by OpenAI. David Barnard (00:37:20): I Jacob Eiting (00:37:20): Figure David Barnard (00:37:20): Just shared data that the Claude app went from making like hundreds of thousands a month to- Jacob Eiting (00:37:26): A day. It was a day. David Barnard (00:37:27): Oh yeah, a Jacob Eiting (00:37:28): Day David Barnard (00:37:28): To over almost a million dollars a day in revenue. So Jacob Eiting (00:37:31): That just David Barnard (00:37:32): Exploded when it seemed obvious that ChatGPT had won that market. Jacob Eiting (00:37:36): Yeah. I mean, look, consumers are ... I will say the ChatGPT app defined the category, but that's rarely that the category, initial category definer remains. I mean, maybe not rarely, but it's not given. And if nothing else, they create space. So don't count ... There's a lot of great ... You think of Perplexity, Gemini, there's a lot ... These are probably categories that I don't know how many of our listeners are playing in. It's just a handful of companies, but that same mindset can apply to every other niche where it's like, yeah, so- and-so's owned this category for a million years, but if you're a little hungry, you've got some tokens to burn and some ideas, I don't know, go poke a bear. You know what I mean? See what happens. Why not? David Barnard (00:38:18): Yeah. And for a lot of these apps, people are going to use multiple. I have Claude and ChatGPT. Jacob Eiting (00:38:24): Yeah, I have a Claude Max. I have ChatGPT Mac or Codex Max. I have all of them. This depends on the day, you know what I want for which each thing. Gemini, I have a Gemini subscription too. I have them all. David Barnard (00:38:34): We just keep coming back to it. I think there's still so much opportunity and- Jacob Eiting (00:38:38): We should have us someone here, David, where we get on. There's no more opportunity. Sorry, no podcasts. You'll know there's no more opportunity when we've all paper hands this thing and we've all exited to private equity or whatever. That's how you'll know. You won't get a podcast. David Barnard (00:38:53): Oh, man. Jacob Eiting (00:38:54): Humor. It's humor, David. It's jokes. It's fine. David Barnard (00:38:57): So the next thing I did want to talk about is one of the key takeaways from the report. I think it's like page four in the report. Hard paywalls crush freemium on download to paid. Hard paywalls convert five times better than freemium. They do 10.7% download to paid by day 35 versus 2.1% for freemium apps. Huge. Jacob Eiting (00:39:20): No day like the present. You know what I mean?That's my thesis on why that's so much better, but I'm surprised it's that start, honestly. David Barnard (00:39:29): This is a tough one because, I mean, I have a hard paywall on my app. I've talked to a lot of people who are moving to hard paywalls. This really does boil down to a strategy decision of, are you building something that's going to be a category leader where that freemium based is meaningful? Are there advantages to having that freemium base? Are they telling their friends about it? Are there network effects? Are there data effects? Is there a reason to have free users in the app? Jacob Eiting (00:39:59): Yeah. David Barnard (00:40:00): And if not, I mean, a hard paywall does kind of make sense. Yeah. Unless Jacob Eiting (00:40:04): You have a viral thing you're trying to do or you want your brand to be. Yeah. I mean, I think this is not necessarily new. I think it's just become more standard practice. And I would say not doing this is probably becoming the exception. Unless you're just a side project or whatever, then you don't feel comfortable. Don't force yourself to do it. But if you've got a good product, if you believe in it, if it's valuable, you think it's valuable and you're doing spend, like you're doing acquisition, unless you're very, very, very dedicated to it, I think it's worth considering. David Barnard (00:40:41): For the right app though, freemium can be super powerful. One of the minisodes I did was with Michael from Mojo, and their app does more than 50% of revenue not on the first day. And so we'll get to stats in this report, but most apps see 80% of their conversions on the first day, day zero, that people are opening the app. But he sees that as a superpower is that they have a generous free tier. And so they do convert a lot of people on day zero on that first day, but then they're able to convert a ton of people over time. And one of the things that he talked about in that episode is they're doing a paywall on launch for freemium users and they're tasteful about it. They only do it once a week and stuff. But it's like now instead of getting this one shot, you get somebody in and if they retain and keep using the app, you've got a lot of opportunities to keep hitting them with like, "Oh, check out this free feature, premium feature, upgrade here." Jacob Eiting (00:41:39): There's probably, this is spectrum, right? It's not hard paywall or full freemium and you know what I mean? And maybe in a phase, David Barnard (00:41:48): Like you launch hard paywall and then you get the product where you feel good about it and then you switch to freemium. Jacob Eiting (00:41:54): Yeah. Yeah. I think your acquisition strategy should inform this probably more than just our one anecdote and stuff. But yeah, every ... I've been saying on this podcast, David, since we've had a podcast, people undercharge, they don't believe their stuff's worth it. They have imposter syndrome and all these things. They think everybody's going to get mad at them because they didn't give them something for free. Look, always. I was having to debate today about stuff in RevenueCat and us charging for it. And it's like, look, they could just not have it. We can just not make it. Is that better? What do you want? So I think it's a thing that this is like a first time second time founder thing some too, but some people are just super sensitive about it. And also just pricing is so transient. If you try it and people don't like it, you can just roll back. (00:42:41): Nobody's going to know. It's fine, but it's definitely something to try. David Barnard (00:42:45): I wanted to have this conversation and be a little more nuanced because if you go look at the report, it's like, "Oh, hard paywall is obviously the thing to do. It's going to make me five times more money." Well, maybe- Jacob Eiting (00:42:55): Just pay Sosa into Claude and just make more money, make no mistakes. I don't know, that might outperform a human overthinking it, right? Let your bot do the overthinking. But I mean, I think it falls into the same category as everything. David, there's this make it a good report and then there's all the nuance, right? It maybe doesn't fit in the report always, but I don't think there's any ... I think the data is true. And there's just like, this is in life and software, there's no time like the present, right? When somebody's in your app and they're using your app, the cross section of intent and numbers and everything is usually just at its peak. So make as much hay out of it as possible because seven days from now, even if your app is great for them, they're not going to care as much. (00:43:40): So maximize that moment. David Barnard (00:43:42): And that's the thing about freemium is that people actually have to retain for them to open the app and see your paywall again. Jacob Eiting (00:43:50): Yeah, yeah. You have them now. You're going to miss a shot. You're going to lose half of them in a 50% day one retention would be great. So you're going to lose half of them tomorrow. And again, it's like they've already crossed a lot of chasms to get to you. They searched, they had intent, they downloaded, they opened, they went through your onboarding. You're never going to have them as primed and as jazzed about your product as they are right now. So capture it. David Barnard (00:44:15): On the hard versus soft paywall or hard paywall versus freemium, the one thing I was surprised is that retention on the hard paywall was right in line. I mean, a little lower than year one retention for freemium. So I kind of would have guessed an app that forces you into the decision day one and you can't even use the app. I would have assumed human meatbag intuition, like those people will probably churn at a higher rate because you like force them into the decision, but it's pretty equivalent. The P90 retention on freemium is 58% and for hardware paywall it's 54, but the median is 27.7 for freemium and 26.8 for yearly, kind of a wash in the median. So you're not sacrificing retention or ... I mean, each app is going to be different, but in aggregate, people aren't sacrificing retention by doing a hard paywall. Jacob Eiting (00:45:15): It's a little bit non-obvious to me that I wouldn't move it, because usually anytime you change something substantial, it's going to move your mix. But there's probably two forces there. I'm not going to speculate, but there's probably one that increases and one that decreases and they're just balancing each other out would be my guess. But yeah, it's free money. Just do it. It's free money. David Barnard (00:45:37): All right. The next thing I wanted to talk about was the window to when a user is closing. 55% of all three-day trial cancellations happen on day zero. This is another chart that when you look at it, it's just massive concentration, Jacob Eiting (00:45:53): Just David Barnard (00:45:54): Stark that- Jacob Eiting (00:45:55): That's the flip side of the same thing point we just made, right? Which is like, most of the things happened in that first day, including them canceling your trial, right? David Barnard (00:46:05): Yeah. People start that trial and then immediately go turn out. This has been a growing trend. I don't think we shared year over year numbers here, but I believe it's been a continued trend that this goes up every year that more and more Jacob Eiting (00:46:19): People- I take super savviness, right? David Barnard (00:46:20): Yeah. Jacob Eiting (00:46:21): More and more people's- Apple's also been more better about sending out notifications and things, but that makes sense. I mean, good, good. Good. Yeah. I love an auto renewing trial. Auto converting, it's great for app developers. It's seamless for consumers, but also I think if consumers are opting out, it's good too. Let them do it. It's their mind. One David Barnard (00:46:43): Little tactical point. I talked to Anmo from Duolingo in one of the minisodes and he said Duolingo had been experimenting with that trial reminder, kind of the Blinkist trial reminder paywall where it's, "Hey, we'll remind you about the trial so you can cancel it. " Well, now Duolingo experimented with letting people choose what day they want to get the reminder on, which draws attention to, "Hey, we're going to remind you about this free trial." And so they saw a conversion uplift from that. Jacob Eiting (00:47:16): Yeah. I mean, you're getting people to not cancel on date zero. David Barnard (00:47:20): Yeah. Jacob Eiting (00:47:20): It's what you're doing. I think. I don't know. And then you're David Barnard (00:47:24): Reminding them. So Jacob Eiting (00:47:26): Maybe the David Barnard (00:47:27): Notification Jacob Eiting (00:47:27): Gets David Barnard (00:47:28): Lost or whatever, but- Jacob Eiting (00:47:29): Who knows? Who knows? These are consumers buying at willing prices, David. That's interesting. I mean, certainly a good look. I'm still a big fan of the blankets paywall. I think in the reminders and stuff, I think it's a net brand positive and trying to trick people. All David Barnard (00:47:40): Right. Next thing is Google's billion dollar leak. Not a huge fan of that framing. It is what it is. Google Play is a very different platform with different consumers, with different forms of payment, but it is really interesting that nearly a third of all subscription cancellations on Google Play are involuntary billing failures. Jacob Eiting (00:48:02): Yeah. We see this all the time with customers being like, "Why am I getting so many billing issues?" More David Barnard (00:48:08): Than Jacob Eiting (00:48:08): Double IOS. I couldn't tell you exactly the reasoning for that. I don't know if it's just economics of who owns Androids. I don't know if it's the way they do this- Some of this David Barnard (00:48:19): Payment method because Google will let you- Jacob Eiting (00:48:21): Use prepaid cards and stuff like this. Exactly. David Barnard (00:48:25): And so then your balance runs out and Jacob Eiting (00:48:27): Then the billing fails. Yeah. The physics are just very different. There's not an auto recovery path like there is on iOS. I mean, there's just a whole number of reasons. It just compounds the fact that makes Android. So even if you get ... You still see whatever, I don't know what our blend is, but it's 10, 15% Android revenue versus the rest iOS probably in web. But you see why. It's not just a market's not there. It's like also the bucket's leakier, the conversion's harder and stuff. I'm good on Google for putting in a good fight, keeping it going. We need two platforms, but yeah, it answers your question about why people are continuing to increase their iOS concentration from earlier. David Barnard (00:49:04): As I was researching this, Google did share one of their developer pages that folks who turn on grace period, and there's another feature, don't remember it off the top of my head, but you need to be, especially on Android, you need to be thinking about this- Jacob Eiting (00:49:19): Managing this problem. Yeah. Well, I mean, there's also communication, things you can do, like you can support. And we have good hooks for this and RevenueCat to make it easier to do all the things, right? Especially once Android's beyond just experimental revenue for you. David Barnard (00:49:33): And it kind of makes sense. The Apple platform is way more oriented around keeping your credit card up to date. Everything is Apple Pay. You've got your Apple subscription to keep your photo library in the cloud. It feels like as a platform, they do kind of more strongly incentivize you to keep your billing updated, keep money flowing through their systems. Jacob Eiting (00:49:56): It's the whole benefit of having a great retail partner, right? They make the buying experience good. They keep consumers coming and you're willing to pay a fee for it, right? David Barnard (00:50:05): And then speaking of churn on Google Play, we did include an interesting chart about the reasons for cancellation. So there's actually an API that we call and get this data from our customers because Google asks when you cancel in Google Play, why did you cancel? And then we're able to get that data and then aggregate it. And so it's interesting. The top reason is cost related. And it swapped back and forth. I think last year it was maybe not enough usage, but- Jacob Eiting (00:50:33): I mean, our does not the same answer though. You know what I mean? It's not enough usage to justify the cost. They're sort of interrelated reasons, right? So I wouldn't necessarily read. But if you David Barnard (00:50:43): Combine the two, that's a good point. If you combine the two as a value proposition, I'm not using it enough to be worth the money, that's 70% of cancellations. So it makes sense. Jacob Eiting (00:50:55): Economics, people stop paying for stuff they don't want. David Barnard (00:50:58): Technical issues is small. Found a better app is small. So you're more likely to lose a customer because the value prop is not there than you are because they found a different app that solves that need. Jacob Eiting (00:51:10): App developers tweets. Is that on there? No. No? Okay, good. David Barnard (00:51:15): All right. Back to AI apps. So AI apps sell, but they don't stick. Jacob Eiting (00:51:21): So this is interesting. David Barnard (00:51:23): AI, our powered apps generate 41% more revenue per payer, but they churn 30% faster. Year one retention is 21% versus 31%. That's interesting. I mean, it does feel like there have been so many generative AI apps launch and some of them are ... The studio gibbly when that exploded for ChatGPT, but a bunch of other apps jumped on that bandwagon and released apps. I mean, Jacob Eiting (00:51:53): I think it's like most of these have not solved the building stickiness. ChatGPT, I'm going to throw them under the bus a little bit on this. It's like, yeah, they have memory, but I don't feel super invested in my ChatGPT app history, right? I open it, I run a query, it's whatever, and then it's transient. Codex is a little bit different because I'm building my Codex workflows and things like this, but that's a very niche thing. And I think this is one of the things that OpenClaw kind of blew open, which was like, just give this thing more stuff about you, right? Give it more integrations, obviously give it more power is helpful. But even OpenClause, an example, at the end of the day, it's a bunch of integrations and skill files, which are fairly transferable, right? And that's on the agent side. But I think we're just in an emerging market where new products are coming out every three months and people are willing to shift and there's not a huge lock in effect. (00:52:52): Now you've seen, I think of the ChatGPT, they've OpenAI's been pushing their medical thing. They're like, "Bring all your medical history in. " And obviously one, I think it'll be a very useful feature, but then secondly, it serves a purpose for them of like, okay, now if OpenAI has, I have my Epic chart, whatever is linked to it and has all my medical records and all of this stuff, I think you start to build those, but it's going to take a while. It's going to take a while for these products to really figure out what their sticky ... Let's assume we hit some sort of terminal model functionality at some point. They're going to need to innovate on more of the consumer features. David Barnard (00:53:31): I think it also just shows people are willing to try stuff. There have been so many new AI powered apps launching to do this photo Jacob Eiting (00:53:39): Manipulation. Oh yeah, I subdid anything right now if it's novel. New photo thing, new audio thing, whatever. I'll try it. 20 bucks a month, whatever. So I think a sign David Barnard (00:53:48): People are willing to try stuff, but AI apps do need to figure this out for the long haul to get those Jacob Eiting (00:53:53): Sticky use cases. Those are the long tail runs. I don't know. If you're building ... But I think you actually have a better ... If you're doing an applied, you have a very different problem. You like your Strava as an example or a Strava clone or something like this. You're still building a data mode of stuff. And actually, I think it's an interesting ... I think that there's a lot of discussion going on, where's the defensibility in software? Because one, it's cheaper to make software and all this stuff, but there's still a tons of defensibility in that long-term customer relationship. It's like Dropbacks has all my photos. Strava has all my workouts, all of these things, and those can't be vibe coded in a weekend. You're actually in a stronger position than these brand new and also mostly stateless AI tools that ... Yeah, I've had better stickiness for my little personal open call deployment just because I can modify it and I'm building an empire inside of it of things it can do, not so much in my ChatGPTs and things like this. (00:54:49): So yeah, it doesn't surprise me. David Barnard (00:54:51): All right. Another hot take that deserves a little nuance, trials of 17 days or longer convert 70% better than short trials, 42.5% versus 25.5%. But year over year, more apps shifted to three-day trials. So this is one where, and I've talked to Rick about this. I think this is a correlation versus causation kind of thing. Will you get better conversion if you have a longer trial? Maybe it's worth trying, but I don't think you're guaranteed that. And the thing about three-day free trials, the reason people do that is cash flow. You brought that up. Jacob Eiting (00:55:32): You want the money in three days, you don't want the money in 30 days, right? And getting David Barnard (00:55:36): Conversion data faster to compound experiments you're doing on onboarding and paywall and everything else like that. So it's tough because yes, in an ideal state, maybe everybody offered 14 days or 30 days or whatever, and the data shows that those longer trials convert better, but ... Jacob Eiting (00:55:55): But it could still be a net loss. If you're losing that information flow, you're losing that cash flow, you're losing all those things. Even you get to eke out a few points on conversion rates might not be worth it. David Barnard (00:56:06): Back to what we were talking about earlier with the day zero cancellations, I mean, that is probably part of it is that when you have a 14-day free trial, it's like, oh, I mean 17 plus meaning 30-day free trials. It's like there's a little bit more of a like, "Oh, I've got plenty of time to check this out. " And you don't immediately go turn off auto-renew because you're like, "Oh, I'll remember to turn that off." Maybe that's some of it just that, but I think there's probably a bit where the apps that are giving 30-day free trials are different kinds of apps than the apps that are offering three-day free trials. Jacob Eiting (00:56:38): Some selection bias, right? And then just apps too that are ... If you're spending a lot on ads, you're probably less differentiated. I don't know if that's true. It's not because you spend on ads, but if you're in a competitive market, you often need to. I think everything in the SOSA report is worth considering some of those dynamics. We can't fully separate causation and correlation. So it's perfectly plausible, right? Which is why, I mean, going back to nuance, it's like why you should never take one of these single data points and treat it as gospel because it's going to be different for your app. David Barnard (00:57:09): Run your business on it. All right, two quick hits as we wrap up. We shared a really cool word chart is the distribution of paywall call to action language. So we shipped our paywall solution ... Well, we shipped our first paywall solution a couple years ago, but we've been accumulating all this data now of what people are actually using on their paywalls. And so you look at this word chart, continue is still like the biggest. Subscribe is the second biggest, but you got a whole bunch of ... This is a fun chart to just look at for CTAs. Jacob Eiting (00:57:40): What are some good ones? I don't have it in front of me, David. David Barnard (00:57:42): Redeem offer, get premium, get started now, start for free, GoPro. I mean, there's just a ton. Jacob Eiting (00:57:50): I want to look at the long tail. Those are the ones that lots of people use. We should look at the ones that very few people use because this is what we realized. Well, I was reading David Barnard (00:57:56): Some of the little Jacob Eiting (00:57:56): Tiny ones. No, I mean the one-ups. Those are probably the weird ones, but- The one David Barnard (00:58:00): I'm not seeing in ... Oh no, it is in here. It's one of the smallest ones, but try for zero dollars and zero cents. Jacob Eiting (00:58:07): I David Barnard (00:58:07): Talked to the chief product officer at Duolingo on the podcast, and that was released back in, I think, January, and he said that was a meaningful lift for them. And try for free did not convert as well as try for $0 this year. Jacob Eiting (00:58:24): Looks like you got a bug. David Barnard (00:58:25): But anyway, it's just another example of how the 2026 report is better and bigger than ever. It really is fun to see this word chart and look through all these different words and phrases in there. The last one I wanted to touch on was the cancellation timeline for annual subscriptions. This is really cool. And we've shared this in different forms over the years, but I actually really like the visualization here. Jacob Eiting (00:58:46): Yeah, I saw that. Yeah, I agree. This is a good way to look at it. David Barnard (00:58:48): What it shows is for people who ultimately churn, when did they turn off auto renew? And when we first pulled this data years ago, it surprised me, I think we've talked about this on the podcast before, but I always expected that for an annual subscription, you would see a large portion, if not the majority of people who churn, churn right before month 12, right before it renews. But no, 34%, this is across all categories, 34% cancel in that first month and only 11%. So there's a bump because month 11 is 4.7% and it bumps up to 11%. And Jacob Eiting (00:59:26): Then right at the renewal time, people turn out again. But David Barnard (00:59:29): It's not like a U, it's like a J, but like a tiny J curve. I just said always intuition would've expected this to be way higher. But anyways, what this says is that people are turning off auto renew when they see that charge hit the credit card or soon after they're making that decision sooner rather than later. So just another, just like we're talking about with day zero free trials and getting people to retain and things like that, it's like, you got to make your impression soon because Just people aren't just sticking around and making that decision at day 364. Jacob Eiting (01:00:05): I was just jazzed about the visualization. I thought that was finally we figured out a way to make it not look dumb and hard to read. We're renovating over here. Wait for the- David Barnard (01:00:13): Innovation. Jacob Eiting (01:00:14): So the report's going to be even better. No, the problem is this is the ideal and next year we're going to try something worse, but that's okay. David Barnard (01:00:23): But do we even need visualizations in 2026? You're just getting Jacob Eiting (01:00:27): Dumped to report into David Barnard (01:00:28): Cloud code and have your agent tell Jacob Eiting (01:00:32): You what's important from the data. We're supposed to plug this. But yeah, I was just thinking about this. So it's one of the inputs. So we're working on some cool stuff here. We've talked about AI stuff, but we've been hacking on RevenueCat's own agent internally, which I'm workshopping right now. I don't know if we'll announce this, but it's Revenue Intelligence Copilot, Rico. He makes you Ricoh. I'm more excited about that than everybody else on the team. So we'll see how we actually ship it. David Barnard (01:00:54): I Jacob Eiting (01:00:54): Didn't know it David Barnard (01:00:54): Was an acronym and I thought Ricoh is kind of goofy. Jacob Eiting (01:00:58): But no, it's Revenue Intelligence Copilot. It just happens to also mean rich. It's Spanish, which I think is funny. But yeah, we've built a ... I think a lot of people in the community have kind of hacked this together, which is like pulling your RevenueCat data and maybe some other stuff. We want to bring this to a first party feature into the Revenue Cap platform as we figure out what does it mean to be a system of record in the future. And I don't think there's any great ... There's not going to be any vendor our customers have they feel great about that doesn't have well-built agentic interaction built into it. And so we're figuring that out. I'll be honest, we're mostly a JSON and databases company, so this is all new territory for us, but it's not rocket surgery, as they say. And what we've seen this thing produce, I mean, David, you've used it. (01:01:42): It's really fun. Yeah. It's great. It's a fun way to interact with it. It also can just do things that you can ... Everything you can do in RevenueCat is just what the agent can do, but it's just faster. It can click faster. It can pull data faster. It adds a little bit of color, which maybe isn't even the most valuable part. The most valuable part is it can just pull the charts a lot faster than you can. Yeah. David Barnard (01:02:03): So currently we've been using it internally as a Slack bot where you can just ask at Ricoh, how was performance last week? But what's cool is you hooked up last year's report and then you're going to have it Jacob Eiting (01:02:17): Ingest this David Barnard (01:02:17): Year's report. Jacob Eiting (01:02:18): So this is integrated on. David Barnard (01:02:20): Yeah. The reason I wanted to bring this up is because I was going to suggest people just drop the report into a Claude project and then bounce your data in, ask it things you should be thinking about. Use it in a project in Claude. But for RevenueCat customers who enabled this feature once we release it, that works out of the box. You just say, how does this compare? What things am I weak on compared to the report? What things should I be thinking about? And it's pretty smart. Jacob Eiting (01:02:52): Yeah. These models are great. Nothing to do with us, RevenueCloud. It's all just like, we just gave it this stuff. It's actually a good idea. We should note this for ... But we should release a markdown version of the report. That'd be easier for people than dropping just a raw PDF in. But yeah, I'm super excited. I mean, I don't want to plug too much of our book, but we're awakening to how we can make Revenue Cap more AI powered and sparkly. And so Ricoh's kind of this first iteration we're hoping to ship. Ideally, maybe this is out before the podcast drops, but we're still putting some finishing touches on it. But we're also working on how do you egentically interact with paywalls? How do vibe code apps or vibe coding environments create paywalls and stop having to click all the time? Basically, in 2026, we're tired of clicking. (01:03:38): No more clicking. We're typing. No more clicking. Maybe talking, but we're not clicking on things anymore. And then also we have an open claw skill, which you can use on your ... You can use it in Claude as well. We should get it on the Claude Marketplace as well. Yeah. David Barnard (01:03:52): I think we have a ChatGPT plugin now as well. It's so fun internally because people just build this stuff and I don't even know all the stuff we have. Jacob Eiting (01:04:01): Yeah. And we're adapting to this new world and I think we're just getting started. So hopefully that's live by the time this drops. If not, stay tuned. We'll find out six days from now. Sign up for David Barnard (01:04:12): The beta. Jacob Eiting (01:04:14): Yeah, for sure. All David Barnard (01:04:15): Right. Well, I think it's a good place to wrap up. Drop the report into an agent and just have it tell you all the things. Jacob Eiting (01:04:22): Listen to this podcast. You're probably not. All David Barnard (01:04:24): The way to the end. Jacob Eiting (01:04:25): You probably had your agent summarize it and is giving you the digest. But anyway, if you have listened, it's fun. These are one of my favorite ones we do over here, David. David Barnard (01:04:34): 2026 is a weird time. It's a Jacob Eiting (01:04:36): Weird year. The weirdness shall increase until morale improves, I believe. So everybody buckle up. I don't think we've hit peak weird yet. David Barnard (01:04:46): Well, I did want to shout out the team, Peter and Lorelei and everybody else who put in all so much work. It was a huge group effort. Biran and we had a contractor working on a lot of the games. Jacob Eiting (01:04:58): I know things must be going well at RevenueCat when every year I hear less and less about it. So you know what I mean? The first year it was like a whole team effort and now it's still a lot of people involved, but it's operating much more efficiently. David Barnard (01:05:11): Yeah. The first year, it took so much more of the small team that we had. As a percentage of all work done at RevenueCat- Jacob Eiting (01:05:19): We pulled in the ops team and all this other stuff. Yeah. David Barnard (01:05:22): Yeah. And Jacob Eiting (01:05:22): Then David Barnard (01:05:23): Now it's like as we've grown, it was still a massive effort, but as a percentage of all the work getting Jacob Eiting (01:05:27): Done at Revenue Cab, it's much more. So dear listener, we do it for you. We're like investing our finite resources and then putting this report together and I think we're really proud of it. David Barnard (01:05:37): Yeah, it turned out great. And so yeah, you've probably already downloaded the report. Check it out and- Jacob Eiting (01:05:42): We'll see you later. David Barnard (01:05:44): Yeah. 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.