66: How to Make Subscriptions Work for Your App with Dan Burcaw of Nami ML

In an era of subscription overload, apps with subscription revenue models have their work cut out for them acquiring (and retaining) subscribers. That’s why Dan Burcaw founded Nami ML, a machine-learning company focused on helping app developers grow mobile subscription businesses by reducing churn and focusing on the core user experience. Dan joins the show to share his insights on the plight of a subscription-fatigued economy, how data collection and user privacy don’t have to be enemies, and how the app landscape has changed since the launch of the App Store in 2008.

In this episode, you will learn:

• Why the subscription era exists and why subscriptions are difficult for digital product owners to manage
• How subscriptions help freemium apps find their best users
• How on-device machine learning works compared to traditional machine learning
• Why privacy-friendly products are the future
• Examples of mobile-first apps doing subscription models well (and not so well)
• What makes something complex to develop
• How the App Store has evolved since inception and its influence on the app ecosystem

This episode is brought to you by The Jed Mahonis Group, where we make sense of mobile app development with our non-technical approach to building custom mobile software solutions. Learn more at https://jmg.mn.

Recorded February 8, 2021 | Edited by Jordan Daoust | Produced by Jenny Karkowski

Show Notes:

Nami ML

AllTrails App

Episode Transcript:

Tim Bornholdt 0:00
Welcome to Constant Variables, a podcast where we take a non technical look at all things technical. I'm Tim Bornholdt. Let's get nerdy.

Before we jump into this week's episode, I have a quick favor to ask. We are conducting a survey of you, our listeners, to hear your thoughts on the show, and help us plan content and pick guests that matter to you. We also want to know what some of your favorite podcasts are that you listen to. So if you have a minute, please head to constantvariables.co/survey. You could even fill it out while you're listening to this episode. That's constantvariables.co/survey.

Today we are chatting with Dan Burcaw, co-founder and CEO at Nami ML. Dan is a serial entrepreneur who has founded four companies, each on the forefront of a major technology wave. He currently leads Nami ML, a machine learning company focused on helping app developers grow mobile subscription businesses. Dan joins the show to share his insights on the plight of a subscription fatigued economy, how data collection and user privacy don't have to be enemies, and how the app landscape has changed since the launch of the App Store in 2008. So without further ado, here is my interview with Dan Burcaw.

Dan, welcome to the show.

Dan Burcaw 1:32
Hey, thanks for having me.

Tim Bornholdt 1:33
Really excited to have you here. Why don't you kick things off by giving us a little bit of background about yourself? And how that all led into building out Nami?

Dan Burcaw 1:42
Oh, man, where do you start on that? I've been playing with computers for a long time. And I guess anybody that was involved in sort of the early internet, you know, we all found ourselves in 2008, working on a new form of the internet, which was internet on phones. And so, you know, that's skipping a whole lot of kind of learning and building stuff. But when the App Store launched, I had been at Apple, and I decided to leave and start an agency to build apps. And kind of early to that world, I mean, there's lots of agencies out there these days. But I had a hunch that building Objective C software, which was how we were making iPhone apps in the earliest days, wasn't something that a lot of companies were going to have the skill set for. So in that company, we grew it, worked on a lot of really high profile applications over a period of time, companies like JetBlue. And we built the official apps for March Madness Live and a bunch of others that we're really proud of.

And, you know, from that experience, I also founded a company that built a cloud service that every single one of those applications needed, which was around push notification. And nobody knew what push notifications were when they were first introduced by Apple. And we would even have a hard time in those kind of earliest days explaining what a push was. But a lot of apps needed them, and especially the ones we were doing in sports and media, news companies, you know, everybody in that realm was using one of these push notifications.

But what happened through that experience is we built that company and we had the opportunity to sell the company. We ended up at Oracle where our technology became the kind of heart and soul of a product inside of Oracle called the Oracle Marketing Cloud. It's kind of a suite of tools for marketers to be able to market to their end users, kind of across different channels, so things like email, SMS, push notification, display advertising, and so on. And kind of orchestrating that all together so that you're not both getting the email and the SMS and the push. You're sort of getting the message at the channel that makes the most sense. So me and my co founder spent a few years there building that out, the kind of the mobile channel of The Marketing Cloud.

And as our customer base was expanding beyond sports, media entertainment, kind of some of those brands that we had worked with, we saw this pattern that was starting to emerge kind of over and over again with some of these other brands in retail or in ecommerce or in travel. And it really was this notion that they all knew they needed to build an app. That's where their customers expected them to be on the App Store and be on Google Play. So they would invest a huge amount of effort to get there. But then once they shipped, they wouldn't know sort of where to go from there. And so oftentimes what would happen is that they would then sort of get stuck in this world of just sort of chasing downloads. And then, you know, in many cases paying for downloads through a paid acquisition campaigns, but then they would spend just kind of a crazy amount of time really focused on churn. So how do we get our users that have downloaded the app to not abandon the app, delete it from their phone, and so on. And so we just saw this over and over again, where you spent all this energy on an app. And then you ended up being focused on kind of these low quality downloads and trying to retain those users. And depth doesn't feel like a, you know, highly strategic thing, in a way. I mean, what's the objective of the app if the objective is just to kind of like, funnel people in and then try to keep them there, even if they aren't even sure why they have your app to begin with, because they acquired it through sort of an ad, if that makes sense. Coming into an app through an ad is very different from coming into an app because you went on the app store. You searched for something specific, you found something that seems to match the thing you were looking for, match the need, and then you give it a try. Right, that's a very different type of user.

So where kind of Nami came from was we were noticing that folks in the kind of more modern app economy don't have a lot of tools in their arsenal beyond some of these sort of growth oriented acquisition tools to buy downloads, analytics tools to measure, you know, how many downloads are kind of using your app and what your performance is around that. And then some of these messaging products, like we built, that are all about trying to keep people in the product. And beyond that, you know, that's kind of the extent of the ecosystem that we saw. Meanwhile, what was happening was, Apple was starting to encourage developers to adopt a subscription model in their apps for ones that were trying to monetize in app. And subscriptions are very challenging for a variety of reasons that I'm sure we'll talk about. But we noticed that it's the techniques to try to get people to not churn from an app or to acquire a user to come into your experience through one of these ads. None of that was really quite conducive to building this sort of durable subscription revenue stream. And so we saw that there's an opportunity to really build sort of like a subscription first marketing suite. And that's what we're building with Nami. And so it's sort of born out of the experience that we had been part of kind of this ecosystem a bit. And then seeing where the industry is moving towards more subscription based models.

Tim Bornholdt 8:03
Seems to be a theme in your career, too, with being able to kind of see what's coming and see where the problems are, and jumping off and attacking them with a solution that's going to actually help people out. I wonder, you mentioned subscriptions being difficult. And you said, we'll probably get to that. I want to jump on that right away. Because subscriptions are really difficult. But I want to hear from from your standpoint, why. Why are we in this subscription era? Is it just because of Apple? Or are there other market forces acting on it? And then the second part being, what about subscriptions is difficult for developers or product owners to handle?

Dan Burcaw 8:44
Yeah, so definitely other market forces, right? I mean, people over the past period of, I mean, even the Netflix model of the subscription around shipping a physical DVD around, right, that was a subscription model. And so in many ways that might have even been some people's first subscription model. And now we've got subscription boxes, and we've got other subscription services, be it software like, you know, the Adobe Creative Suite for your computer that used to be what a couple $1,000 piece of boxed software that now you can pay a subscription to get access to it. So I don't totally think that Apple was sort of the innovator around subscription. However, the thing where it is sort of a Apple driving the industry is that a lot of these applications were either, you know, free and not monetizing at all or free with a one time in app purchase, which can work really well in kind of a game context where you need to keep buying coins or ammo or something to kind of keep going in the game, but doesn't really work very well outside of the game category. Or there's been a lot of ad based monetization, which, you know, we all know the ads are sort of a lousy experience, especially in mobile, where we've got these limited screen sizes. And so that's been sort of a lose lose model. So I think Apple kind of pushing people that direction at publishers in particular, is to create an environment where the app publisher can have more of a durable, as I say, sort of a more of a recurring revenue stream. So they can keep reinvesting in their apps.

But from the end user perspective, what it does is it kind of segments your user base a little bit more where, you know, so let's say you're a freemium app, and you have some free functionality. And then now you have a subscription offering on top that unlocks some additional premium features. So subscription is great, because it gives you an opportunity to segment your users between folks that are sort of casual with what you're offering and folks that are sort of, you know, your best users, your best sort of fans. And so it's just a different mentality, because the subscription of any given app, it's not meant for every single user of that app, necessarily. So it's just a different mindset.

And then on the technical side, just the nature of a subscription that recurs over time. So you know, maybe you have a monthly plan, or maybe you have a yearly plan. So what happens is that, you know, you're gonna get charged for renewal when that month is up or when that year is up. And so there's a time series element to a subscription that means that you're kind of technical infrastructure, your plumbing behind building an offering, the subscription has to be aware of kind of the context of the user. Is the user in their free trial? Are they in their first renewal? Or are they in their second renewal? If they're in their second renewal, and they have disabled auto renew, that indicates that they might turn out of the subscription. So there's all this state that exists in a quote, unquote, a subscription, that is different for each and every user. So you have to manage your users accordingly. There's a lot of implications from the technology perspective. And then there's a lot of mindset around how you monitor those signals to make sure that you're reacting appropriately. The user is turned out of the experience, well, they should no longer have access to the premium content, or if they have turned off auto renew, but they're still an active subscriber, what are you going to do to try to win them back?

Tim Bornholdt 12:43
Yeah, that's why I chuckled when you had said that subscriptions are hard, because that's what's gone through my head as a developer is all those little use cases. When you, like you said, Netflix, take that, for example. That's a monthly recurring subscription that Netflix doesn't really have a free tier. So it's a lot easier to conceptualize, you pay us, you get access, and when you're done, then you don't get access anymore. But if you have that kind of freemium model, where you offer up certain parts of your app for free, and then certain parts of your app are behind a paywall of some sort, it's more like branches of the tree of logic behind your app that you have to account for with that subscription. And so that's what makes it like, from a development standpoint, difficult. But maybe you could also speak too from a consumer standpoint. I mean, ever since, you know, maybe two years ago, when subscriptions were being really pushed by Apple specifically, it seems like every app now wants me to pay, you know, 2.99 a month to get access to them. And there's some apps that I do subscribe to on a yearly basis, because you just want to support certain developers, but I mean, do you see a lot of fatigue going on as well around? You know, what's the threshold that consumers have for subscribing to, you know, 1000 different apps at the same time?

Dan Burcaw 14:06
Oh, they don't subscribe to 1000 different apps. So that's the thing is, you know, if I look at what I subscribe to, as an example, it's probably a set of fewer than a dozen or two, and it's things that I'm into. So yeah, I've got some of the video streaming services, of course, you know. Duolingo is a great example, language learning, right? They've been on the App Store from, you know, early on, and then they've built this really, you know, massive platform now around helping people learn languages. And it's so robust that if that's something you're interested in, you're willing to pay the subscription. So that's sort of what I was getting at earlier around the idea of finding your best users is that not every app, in every subscription within every app, is a match for every single user on the app store, and that's okay. Now, just because a subscription is offered to you, you know, maybe that's fatiguing in the sense that people are moving to the model. But ultimately, the consumer gets to decide, Oh, yeah, this makes sense for me, based upon what I'm interested in, or, Nope, you know, I'm looking for like... Sometimes there's little utility apps that maybe couldn't justify a whole, you know, even a low price, low price subscription, because there's just not enough functionality there. Well, maybe that's a product market fit issue more than it is about subscription fatigue.

Tim Bornholdt 15:28
Yeah, that makes sense. Moving to a different topic here, you know, Nami uses on device machine learning, and I've personally been obsessed with machine learning for a while now. Would you be willing to give my audience here a brief explanation of how on device machine learning works, and how that might differ from other types of machine learning?

Dan Burcaw 15:50
Yeah, yeah, maybe the first context that I'll share is why are we using that in our product. So there's three things that our product fundamentally does. Number one is it simplifies the entire kind of subscription transaction piece. So that kind of hard work that I talked about earlier, kind of managing the lifecycle of a subscriber and kind of understanding where people are so you can grant access or not grant access, or try to win them back, we just eliminate all that hassle and do that on your behalf. So sort of like how nobody really builds out their own push notification service anymore, because there's push notification providers. We're kind of providing that same engine for subscriptions.

The second piece is that we build or we offer a purchase experience. And so you mentioned those paywalls. So our product lets you build and manage paywalls from the cloud. Now, there are native components that live inside the app. So they feel natural to the app experience. But a marketer kind of a non technical person can use our platform to make changes to the marketing copy or the screenshots, not the screenshots, the artwork, kind of like the App Store storefronts. You don't need your developer team to go into the storefront and change the screenshots. You just go in there and make that change. And so we're trying to extend that kind of flexibility to the purchase experience.

And then thirdly, we've got capabilities around sort of analytics and optimization so that once you're selling subscriptions, you can figure out how to take it to the next level and improve and make sure that you're offering the right subscription and the right price point, seeing how changes being made of that purchase experience are having a positive impact. So that's sort of the three components.

But underlying all of that is this on device machine learning piece. And so first and foremost, where this first came to sort of the consumer side is iPhone 10 comes out, and Apple launches face ID. So you now can unlock the phone by the phone just sort of looking at your face and painting this point cloud and running it through a fancy algorithm that says, Yep, that's Tim, let's go ahead and unlock the phone. What's happening there behind the scenes is actually not going up to Apple's cloud. So there's not, you know, your headshot or your picture going up to Apple. That processing is done on device, so that it's really, really fast to unlock. But it's really much better for your privacy. So that's one of the applications that you know, we use every day. And there's others, especially in photography. There's a lot of applications around this on device machine learning. And so it's really about what you're trying to do is shift the processing of some advanced algorithms to running resonant on the phone itself. So that it's fast, and so that it's really privacy centric. But why would you even want to use an algorithm at all? In some cases, it's there's an experience that you're trying to deliver that kind of a traditional computing model where you're just programmatically building something isn't going to work. So like, you would never build a face unlock technology that was specifically coded to detect your face. You need machine learning to train on, you know, that what a face looks like, in a general manner, and then be able to recognize specific features that relate to you, Tim. So it's a generalized way to build something like face ID.

So the way that we're using it in our product is, I mentioned earlier that a lot of app publishers spend a lot of time on this churn problem. Similarly, if it's a freemium experience, some publishers spent a lot of time on kind of the conversion side of the problem, so that the person has been acquired. So they're in the app, but they haven't yet taken you up on the in app purchase, the upsell. So what we wanted to do was build a set of algorithms that could detect if somebody had a propensity to purchase in the app, purchase your subscription, so are they likely to subscribe based upon patterns of behavior that they're exhibiting? Or are they showing early risk of possible churn? So maybe they've not turned yet. But they're there, their usage patterns are showing that they're kind of trending in that direction. And providing our customers with this, that signal, as triggers for potential marketing is really powerful without having to kind of learn that from a more cloud centric model, where you're sweeping up all this data, you're sending it to your cloud somewhere, you're finding these patterns eventually. And then what happens? Well, those patterns end up being used to build an audience segment, to send out another email blast, to try to get the user to, you know, upsell, or to not churn. So what we wanted to do was cut down on emails and cut down on push notifications, if we could give our customers the ability to know about some of these signals while the user is still in the app. What might you do with that? So if somebody is showing a high propensity to purchase, let's say, you know, they're a 9 out of 10 on the on the scale of likely to subscribe to your subscription. Well, maybe that then let's prompt them the paywall, because they're actually likely to subscribe versus if you're a 2 out of 10, maybe we don't show you the paywall, and we let you keep using the free experience.

Tim Bornholdt 21:36
That's really interesting, like the prevailing wisdom, like you said, is the exact opposite of why don't we just take all this data and throw it up on the cloud? It's really interesting to hear how you're doing it on the device. And in most of the contexts that I've seen it in so far, like you identify it, and then like in photography, that's like kind of the big one that everyone talks about, like facial recognition, and things like that. One question I have for you is, if you're doing everything on device, how do you incorporate things that you've learned with one person's behavior on, you know, device A versus somebody else's behavior on device B? Is there some way that you're kind of bridging all the data that's computed on device, but then bringing it up to the cloud, so that you can analyze it later? Or is it really all just truly kind of learning all individually as they go on their own individual devices?

Dan Burcaw 22:28
Yeah, so excellent question. So in today's, we're all talking about today's world and kind of where we're heading. In today's world, we do what's called feature engineering on the device. And so that's the sort of what are the inputs to the model that's going to yield the answer we're looking for. So the inputs are, you know, x one, x two, x three, and we want to get answer why those inputs are piping into the model on the device. We get the answer on the device. And so we're determining what those x's are, those inputs on the device itself. Whereas a lot of traditional machine learning that's out there and in the field, and you know, is used in commerce, you sweep up all this data, you put it in front of a data scientist, and they comb through it to try to figure out what those inputs are by collecting everything. And so you're sort of kind of collecting everything, and then looking at it later to kind of figure out what's important. So that's easier in some ways, because you just, you just grab everything, but you don't know what's important. And then you figure it out, when you have time to figure it out. We are not doing that. We have a subset of things that we're looking for. We're keeping much of that on the device itself. And how were we able to do that, part of how we were able to do that is that our CTO has built models like this in the past. And so he knows just from real world experience, what inputs matter, and that he's built in the cloud model. So now it's taking some of those learnings and applying it to an on device world. And there's all sorts of things that you would, I mean, I think that there's also this conventional wisdom out there that that if you collect everything, like there's a lot of hunch work that happens. So I wonder if this particular type of information might drive the model that I'm trying to build? Oh, what about this other thing? What about this other thing? And so you start to kind of create this grab bag, and then you find out that actually those things don't really move the needle the way that you thought they did. And so having that role practitioner I think is really, really important. And so that's how we've been able to kind of skip the step of just capturing everything, which is good for privacy as well, and we can talk more about why we care a lot about privacy. So that's kind of the one part of it is doing that feature engineering on the device.

However, in today's world, we can't actually do the training on the device. So we still have to send up some data to our cloud, train the models and push the models back down to the device to actually operate. And so the data that we're collecting is anonymized. And it's a minimized set of data. We want to get to a place where we actually don't need anything in our cloud. And we think as these devices get more powerful, and companies like Apple keep moving the needle, they'll allow you to take a generic model that is on, you know, a device one, same generic model on device two, and sort of adapt each of those independently on device one and device two kind of as the users using the app without any data whatsoever going back to the cloud for training. But we're not there yet. Both in terms of the compute capability of the devices, and as well, as you know, is that even something Apple wants too. I think Apple probably will allow that, because it'll help the privacy footprint even more. But we're just not quite there yet.

Tim Bornholdt 26:16
Yeah, it's early days with all this technology, which is what makes it super exciting to be a part of and being on the leading edge like you are. It's really fascinating to see all the different approaches that people are taking to build these models that then translate to actually doing the machine learning, so to speak. You had mentioned before talking about user privacy, and that's a theme that's, I didn't set out to build a podcast that just rants about user privacy. But I think it's kind of slowly devolved into that because there's certain like, points that you can drive home 1000 times, but you'd like to get them from different angles. And I think when people talk about big tech and the problems with it, it's really hard for non technical people to understand why having all this data is scary and bad in the hands of like one corporation that just, you know, you suck all your data into a black box, and you don't know what they're doing with it. But then all of a sudden, everywhere you look, you're getting ads for the same thing everywhere, and you know, so on and so forth. So you'd mentioned that your company really cares about user privacy. I'd love to hear why there's such a focus on that, from your standpoint.

Dan Burcaw 27:29
First of all, we're users ourselves. And, you know, so we're thinking about how, like, if we're gonna be users of apps that offer subscriptions that are powered by our platform, like how do we want to be treated in an ideal world, to try to build to that? And just also just seing, you know, the tea leaves, I mean, more regulation around this stuff. You know, certainly the example you talked about of banners and different advertisements kind of following you all throughout the internet. I mean, at first, it kind of seems like, Okay, well, I'm seeing that same, you know, Nike shoes ad over and over again, everywhere I go. And that seems sort of, you know, benign, somewhat, but then, you know, as people start to be savvy about algorithms, and the kind of social networking space and kind of, you know, how that's shaping people's views, you know, this information is very powerful.

And so we wanted to think about how we wanted our data, you know, as users to to be treated, number one. And then the other thing is, is being inside of a kind of a large organization that has a lot of assets around marketing and kind of advertising technology, to actually see somewhat how the sausage is made. It's just, we just decided that as a core value when we started this company, if we could build it in a more privacy friendly way we want it to. And then I say the third thing back to the tea leaves is that just like Apple has been pushing subscriptions, they've also been getting more and more focused, I would say, about telling a privacy story, you know.

In the early days, well, I say the early days, you know, in the last decade, when Apple would introduce a new Mac, let's say they would, at the end of the presentation, do their kind of environmental checklist where they would tell you, you know, about all the materials that were were used and the recyclability. And I've been talking for a while about, you know, they really need a privacy checklist when they do some of these product introductions, because they've got a great story there compared to some of their peers. And the more they talk about it, the more consumers will be savvy to the fact that, Hey, there actually is a choice around some of this stuff. If you want a smart speaker, there's Alexa, there's the Google speaker, there's the home pod, right, and there's different traits, if privacy is something you care about. There are different traits between those different devices. And so I just have been watching as they've been talking about it more and more. Now with these privacy nutrition labels for apps, right, where this is the future, where things are going to become more and more privacy focused, if not privacy first. So I think you're going to see a whole next generation of all kinds of tools, be it, you know, web analytics tools, or, you know, even just consumer tools that start to emphasize the privacy characteristics basically as a benefit of the product.

Tim Bornholdt 30:33
I couldn't agree more. And it's funny you mentioned the speakers, I'm staring at my home pod right now in my office. We just invested, my family, into putting a home pod in the kitchen. And then we got one of the bigger ones down in the basement where our Apple TV is plugged into, and I've been experimenting with them for the last couple of weeks. And everyone knows that Siri certainly has its quirks in terms of responsiveness, and accuracy, and all of that. And as somebody that understands how all of this works, it's just fascinating how they're able to use machine learning, you know, on Siri and try to do everything they can to keep stuff on device and kind of evolve their models and algorithms and things in a private way. As opposed to you know, yeah, obviously Alexa and the Google Home assistant, whatever it's called, you know, those work really well, much more accurate and better when you see those videos of like, somebody's asking Siri a question and someone asking Alexa a question. It's like, sometimes you're just like, what the hell, like, What is wrong with you, Siri? But on the other hand, as somebody that understands the the privacy behind it, you're like, you know, I'll take that trade off of knowing that my, you know, maybe bad question or, you know, ignorant thing that I'm asking of my smart speaker, it might be a little slower. And it might be a little, not as accurate, but at the very least, it's not being swept up into this vast vacuum of information that will later be used against me, in one way or the other, as I'm going about my business on the internet.

Dan Burcaw 32:13
Yeah, and it would be one thing if it was very privacy focused and wasn't getting any better. But it's been getting a lot better. And so, you know, it's just sort of like the Apple Maps thing. People still talk about Apple Maps. I mean, I haven't had a complaint about Apple Maps in quite a while. Now, I'm not saying that there's not complaints out there, but compared to where it started, and where it's at now. And I think the same is true for Siri. And so if again, you kind of, you know, put privacy in one of your top, you know, three or five things that you care about, then the decision is pretty clear.

Tim Bornholdt 32:44
Yeah, I totally agree. I agree with you about Apple Maps, too. I mean, maybe a year ago or two, I decided to delete Google Maps off my phone and just use Apple Maps just to see how it would work. And they've been making so many improvements to it. And really investing in that technology. And it's kind of like table stakes now because of how good Android is with all the Google services. But it's interesting how you can have two different approaches that essentially get you to the same place. But one is done on a model based off of, you know, kind of respecting user privacy. And the other one is built not necessarily, like, you can trust Google, maybe in some ways, but you know, a lot of people probably don't. And it's because you don't really know. It's a black vacuum that they've kind of throw things into. And I don't know, I'm kind of rambling here.

Dan Burcaw 33:35
I think what the other thing is, these are architecture decisions. And once you make the decision, it's sort of hard to cross over to the other architectures. So that's, you know, if you have a worldview that privacy is going to become an even more important, you know, sort of trade in products going forward, and maybe going forward is not at a time horizon of two years, maybe that's in 10 or 20 years. But if you just think that that's likely the case that privacy will become more important, then why wouldn't you want to architect for that? Versus if you architected it for expediency to build sort of something faster, that's more accurate, faster, you're gonna have a much harder time when it's time to cut over to that privacy forward path.

Tim Bornholdt 34:18
Right. I mean, if you can make one argument about if you are able to innovate quickly, and find those points for your model that matter, and, you know, so on and so forth. And maybe there's a way Google can pivot or or Amazon for that matter, but yeah, I don't know. It seems if you're focused on privacy from the get go, then you just don't really have that concern to really deal with. One question. We were talking about subscriptions before, obviously. And one question I had, maybe bringing some tangible advice to our audience here, obviously your company has a lot of companies that you work with. So you've probably seen things that work well when it comes to subscriptions and things that don't necessarily work so well. What apps out there are you seeing that are actually doing a good job of scaling their business with subscription services? And you can obviously hit Netflix and we've gone down that route. But if there's any other kind of apps that you think out there are good examples of people that are doing it right, I'd love to hear them.

Dan Burcaw 35:20
Yeah. I mean, there's so many kind of mobile first things that are doing a really good job. I mean, here's one I mentioned, Duolingo, earlier. But here's another one that might sort of surprise you. Have you ever used the app Alltrails?

Tim Bornholdt 35:33
No, I haven't.

Dan Burcaw 35:34
Okay, so Alltrails, it's been around for a few years. And it's basically a hiking app. So they have sort of this massive database of different hikes that you can go on around you. And I think they do some more now. But that's sort of where they started. It's just like different trails that exist on the planet. And they were that sort of the index in a product for that. And given the fact that if you know, if you want to go on a hike, you're out in the world, you've got your phone on you, it's not like you're lugging your computer around or a map in the old days, then this is the resource. And they built a pretty large, over time, a pretty large audience from kind of a number of monthly active users. But I just saw the other day, and they've raised by the way, it's like just a ton of VC money, which is usually indicative that the audience is growing pretty fast. But I just saw the other day, I think it was on LinkedIn that their CEO posted that they just crossed 1 million paid subscribers. And that just blew me away. Because if you think about, you know, does everybody in the world hike? No. But there's a lot of people that hike and evidently there's a million people that are willing to pay money every month to get access to this kind of great database of hiking trails. So it sort of seems niche on some level, but then when you start to think about it more, it's like, well, it's actually a pretty large niche, and million's probably just the beginning. So they've done a really great job as well, their experience is really good. That's the other thing that we see a lot is that if you've thought about your subscription as sort of a part of the experience, and what I mean by that is like, not disjointed. So sometimes there's apps that you'll use it and kind of you understand what the apps about, and then they pop up the paywall, and it sort of like just feels like this bolt on component that isn't consistent with what the experience was up to that moment. And then you dismiss it. And now you're back in that kind of other experience again. The best apps, it's a very consistent and seamless experience. And so the upsell opportunity just feels natural and reinforces your everything that you've experienced thus far and kind of your emotions behind, Yeah, this is a great app. If it feels like a bolt on, well, why would anybody want to take you up on that?

Tim Bornholdt 38:02
I think that's a really astute observation. Because there's apps that I've been using since the dawn of the App Store. One that comes to mind is an app called run meter. It was one of the first running apps that came out that used your GPS to track you while you are running, which now seems like, Well, duh. But back then it was like, Oh, man, sweet, I can finally do this. And I've been using that app ever since it came out. And they came out with this elite version of the app. And basically they've added on like Apple Watch components and integration with, you know, Strava, and all the other kind of things that are out there. And it's interesting, because I think if they would have started from scratch, and built it with that kind of mindset in mind, then it wouldn't be super disjointed, but it's kind of hard to pivot into a subscription model if your original model was, you know, premium or something else where you paid upfront for it. And then, you know, you can unlock Apple Watch support for, you know, $2 in app purchase or something. Can you think of any examples of apps that have like, pivoted into a subscription model successfully?

Dan Burcaw 39:18
Some of the fitness ones clearly have. But even now that I think about it, I mean, a lot of the fitness apps were sort of free with ad support early on, and then maybe a couple of in app purchases. So you know, that's one of the categories really doing well now with subscription, but if you think about it, you know, a lot of these are relatively new properties. So I'm not sure if they pivoted. I mean, the thing I've seen, I'm trying to think of an example that is that I can name, but I have seen these sort of, you know, next generation or sort of like so what you were talking about with run meter. Had they sort of decided, okay, well now we're going to build run meter Pro and kind of treated it as a new generation of their app and kind of, you know, build it from the ground up around that. So rather than just trying to like shoehorn subscription on top of something that they built, thinking about it more holistically around, well, you know, is it the Apple Watch support thing that's gonna drive people to want the subscription? Or is it, you know, these other things? And one mistake or kind of issue we see regularly is that sometimes folks look at what others are doing well, or what they think others are doing well, and copying. And this isn't just a subscription, it's, you know, down to App Store keywords, down to, you know, icon design. I mean, just looking in your category, and kind of looking at what your peer set is doing. But what I always tell people is that, you know, just because somebody looks like they're doing really, really well, chances are, and maybe they are, but chances are the decisions they made that came together with a set of packaging, a set of features, a set of price points, a set of sort of the whole of all of those individual parts. There's underlying context and decisions that were made to arrive at what they arrived at. And so if you just kind of copy it and say, Okay, well, they're charging 9.99 a month, so we're gonna charge 9.99 a month. They have Apple Watch support in their pro package. Okay, I'll put up a watch support in pro package. You might be missing why they made some of those decisions, which makes copying not necessarily as efficient in my opinion.

Tim Bornholdt 41:39
Yeah, I'm fond of saying that you should borrow inspiration, not straight copy, but borrow inspiration is a nicer way of looking at other apps and seeing what they're doing and copying it. But it's not necessarily just making a blanket copy of somebody else's product, because you really have to think about it holistically and from the beginning, otherwise, it's possible that you know, that they kind of backwards, you know, walked into whatever model that they've got, and they're not happy with it, and they're not making any money. And, you know, you could also just fall prey to the same issue. So I think, regardless of whether you go with a subscription model, or any kind of model you go with with your business, it's really thinking about, like you said, putting the user experience first and trying to think what is actually going to drive somebody to, you know, click the subscribe button on your, you know, 9.99 a year subscription, or whether they're going to buy it again and purchase, or whatever it might be. It's just really getting in that mindset of the user and optimizing for that.

Dan Burcaw 42:45
Yeah. And also understand, what are you trying to achieve? So some of these mobile first properties, like alltrails, I mean, there's not a download on them, you know, maybe there is. But there's, you know, it's not a computer oriented experience, or it's not a Roku oriented experience. It's a phone app, that's what they are, they're mobile first. And so for them, you know, they've had to find a way to monetize this massive audience that they've put together. And there's really two choices. You can be ads, or it can be subscription. And that's kind of it, or some b2b thing, I guess, kind of, you know, on the side, or b2b to see some sort of partner integration deal, I guess. And so once you make a decision of like, Okay, well, this is what we're trying to accomplish. And then the question is, well, how do we get there, and one of the major issues that I still see just really across the board is that, app developers and publishers still just don't have enough agility to experiment and try things and learn. So that's back to that thing I was saying earlier about, you know, if you're just copying without understanding the context, then you're not really learning. You're just sort of, you know, you're just at it, you've arrived at a starting point for you without really knowing why you're even there other than somebody else was doing it. So if you have agility, if you can iterate quickly, if you can, you know, use our solution to modify your purchase experience dynamically, so you can see what's working and what's not working as just one example, then you can build learnings, you can try new things. If something's not working, you can change it quickly. You don't have to run through another dev cycle. And so sometimes there's one example I like to talk about sometimes is there's a fitness app, I forget if it's fit plan, or eat fit, one of the two, I think. And for a while, they, and maybe they still do, if you go into their freemium experience, which not all the fitness apps do, a lot of them are you have to start the free trial. But these two have a freemium experience. And so if you've bypassed that initial kind of subscriber, you know, do you want to start the trial? You've said no. So now you're in the free experience. And then you're navigating around, you're doing things and they pop up this sort of time sensitive paywall, kind of mini paywall that says, Hey, in the next 10 minutes, if you start the trial, you know, we'll give you an intro pricing. And it's pretty neat. But I can also put my engineering hat on and say, Man, that probably took a lot of effort to engineer that particular component. So if you were introducing a new fitness app into the ecosystem, and you saw that, and you thought, Oh, well, they must be doing really well. And it must be because they've got that countdown paywall. And then you go say, Well, we need that. And then you go talk to engineering and engineering is like, Well, that's gonna take us two months to build. And then you're like, Huh, do I really have conviction that that's the thing that's going to be the difference between whether we're successful or not? And that's what I'm kind of talking about with agility is if you don't have the ability to move quickly and learn, then any one of these kind of point decisions in a moment of time doesn't really add up towards kind of a trajectory towards what you're trying to accomplish.

Tim Bornholdt 46:22
Yeah, taking it as a scattershot approach doesn't really help a whole lot in this space, especially considering how expensive it is to engineer something. It's one thing to, you know, whip up a Photoshop mock up of an experience and test that. But it's another thing to really sink those development hours into it. And yes, I laughed out loud when you were talking about the engineering effort to put up a prompt like that with the time sensitive countdown and everything. A lot of times, I always go back to this XKCD comic where they're talking about, you know, I'd like to build an app where it shows all the birds in a park and the person's like, Yeah, of course, you know, there's a GIS database I can pull from, you know, that shouldn't take long. And then they're like, I'd also like to have the app categorize each bird as I take a picture of it. And they're like, Well, that'll take a research team and about two years to come up with the answer to that. That's one thing, maybe you have some insight into that of, how do you, this is like way off of topic of what we're talking about, but it's kind of an interesting thing that I want to bring up from time to time is, how do you explain to people, when you're talking about those engineering efforts, what it takes to kind of make those pivots? How do you help people understand what makes something complex to build and simple to build, if that makes sense?

Dan Burcaw 47:53
I mean, I guess the tests for me, we even have this inside the company, you know, when we run a product meeting or we do a weekly kickoff where we kind of talk about the engineering priorities for the week. And, you know, there's that thing that happens, where the thing that you think you're gonna, you know, that first thing, Oh, let's go build blank. And you know, that is simple, maybe it's a little nice unit of work. And it seems like you can get it done pretty quickly. But then the next thing that happens is, and this even happened this morning on our kickoff was, somebody will say, Well, you know, but it also needs to do blank. Just like you said, with the bird example. And even that little ask seems simple. And then there's another little ask, and another little ask, and another little ask, and suddenly, you know, it's not that you've necessarily even doubled or tripled the scope. But you've just taken a unit of work that's pretty simple, and you've made it more complex just to even kind of grow. And especially if you don't have clarity around what are you even trying to achieve. Because sometimes those little asks aren't in service of the goal. They're in service of something else, like, Oh, well, we need to be, you know, like, the item this morning was what we need to make sure we're doing the right thing from an accessibility perspective. Well, yeah, that's true. We do want to do the right thing from an accessibility perspective. But we also don't want to like, we have to constantly try to find that line of shipping regularly, you know, units of work, to get them in front of customers to get feedback, because you need that feedback loop and polishing it to death, and shipping rarely, and then losing out on all of that time, time you need, or all that time that you would, you know, would want to have that customer feedback that you missed out is sort of a missed opportunity of feedback, because you polished it too much.

Tim Bornholdt 49:54
So really, it's finding a way to help people understand when you're working on a technical project, as with any project, it's keeping things simple and clearly defined and finding ways to get, even if it's a basic feature in people's hands and get feedback that's, you know, positive and saying, Hey, you're in the right direction. It'd be also cool if it did X, Y, and Z, as opposed to trying to envision what x, y and z are from the get go. Let your customers actually tell you what those, you know, variables might be.

Dan Burcaw 50:26
Yeah, I mean, and in service of a goal. And if the goal is, you know, to be the kitchen, or the Swiss Army Knife of kitchens, well, then I guess you need all of the knives. But most products don't start off their life saying, we want to be, you know, all things to all people. It's they kind of end up that way because of lack of focus.

Tim Bornholdt 50:51
Yeah, that makes sense. Well, I want to leave this off with you've been in the App Store, obviously, you've been helping to shape the App Store since 2008. And you kind of ran down some of your clients that you worked with before with JetBlue and March Madness Live. And I mean, the list goes on and on. I'm curious to hear from your perspective, how have things changed in the App Store since 2008 until now? What are some of the things that you've kind of looked back on now and think, man, that's really cool?

Dan Burcaw 51:23
It started out so simple, in a way. I mean, back to kind of related to what we just talked about, I mean, it was you build something, you chip it on the phone. And that was a big win. And now with all the form factors, and, you know, watch OS and TV OS and the pad and then, you know, the SDK itself has all these bells and whistles, and so you have to be a little more choosy about, Well, hey, just because there's AR kit, do we really need to take advantage of that? Oh, just because there's core ml in on device machine learning, sounds cool, like is there actually value to us taking advantage of that technology? So I do think that we're farther down field from where we started. And that shows very much in terms of the maturity of the ecosystem and the complexity of the ecosystem. So it kind of makes it a little harder to kind of build, certainly more complex.

The other thing is, in the early days, I think we were, you know, a lot more focused on Well, we've got to have an app, right. So if I, so not everybody was that way, but a lot of the brands that I worked with, they knew they needed to be on the App Store. And that was kind of the goal, right? So once they got on the App Store, there wasn't that next goal to say, Okay, well, now that we're there, let's try to meet this objective. But guess what, eventually, you have to have more goals. And so then the next one that would show up is, Well, let's try to have the best app ratings that we can, you know, let's have at least a four star rating, or let's try to strive for a five star rating, or let's try to get featured by Apple, right. So then these goalposts would kind of change.

And so now, though, fast forward, I think that while those things are still important, there's so much effort still focused on making sure that the app reviews and the star rating is as good as it can possibly be. And people still want to be featured by Apple, of course. But part of the maturity in part of the fact that these are more complex objects now deployed in lots of different form factors is that, you know, there's serious money going in. And so that's serious money, serious resources, serious time. One of the brand apps that we worked on, you know, it was for a sports league, and not the March Madness that I talked about. It was for another sports league. And, you know, initially we would work for kind of, you know, their season would kick off in the fall, and then the their kind of championship was in the summer. And so, you know, we would kind of kick off the work in the fall, and we kind of ended the work after playoffs, and then so it wasn't like a year round of effort in a way because their season wasn't a year round. But as this got more complex over time, with all the different devices and form factors, and technologies, it just turned into a year round of work, even though the seasons not going on all the time. And so what I think that that has also resulted in is, is much more and this relates back to subscription, again, much more of a desire to try to figure out what is like, why are we on the app store? Are we on the App Store because we're trying to create a revenue stream? Are we on the App Store, and when I say App Store, of course, I'm using that generically. You know, there's lots of other ecosystems, you might want to be on the Roku store as well. But, you know, are we on there because our brand needs to be there? Are we on there because we're trying to make money directly through something like a subscription? Are we there because we need to offer a thin client into a kind of broader solution set that we have, kind of, you know, through our website, you know? Why are we there? And to kind of justify the investment that's necessary because it's just much more, it's much more work these days. There's no two ways about it.

Tim Bornholdt 55:17
Yeah, I couldn't agree more. I really like your point of, Why are we here? I think that's like one of the most astute questions to ask whenever someone comes to us and approaches us for getting worked on. That's almost always the first question I ask. It seems naive, but, I mean, look at the early days of the App Store. I remember I stood in line for the first iPhone. And then when the App Store came out, I think I downloaded almost every single app. At least you were able to go through and look at every single app. And there were some with clear, you know, purposes, like Pandora made perfect sense to have an app, AOL Instant Messenger, when that was a thing. Yeah, it made perfect sense to have my buddy list right there on my phone. But as time has gone on, every single company has an app now. And when you walk into any retail store, for example, it's always, download our app, get our app today. And it's kind of like, Why do I need an app for this or that or the other thing? Just thinking as a user. So it's interesting from a business standpoint to think about what benefits would people have to actually take the time out of their day to download your app and then maybe use it? It's one of those questions, again, that just it bears frequent repeating, because it doesn't make sense. It makes sense to some regard, ratings and reviews certainly do matter and make a difference. But just because you have a five star rating on your app doesn't mean that people at the end of the day get value out of it. You just have to keep coming back to the drawing board, you know, time after time, season after season, like you said, and continue to find ways to innovate. And Apple and Google are really good at every year putting out new capabilities and giving you new, you know, sensors or new device capabilities that you can then unleash into your own app. But again, you have to keep approaching it of, Do I really need machine learning in my fart app? Or like, what's the purpose of all this stuff?

Dan Burcaw 57:29
Well, you know, the one category that is sort of interesting is the casual restaurants. I know, remember, a few years ago, it's like there was a 711 app. And then there was a McDonald's app. And some of these, you know, fast food sort of destinations. Starbucks has been on the store for a while. So that's a little bit different. But you know, some of these users here are like, Do they really need an app? And that was sort of my reaction with some of those. But what's interesting is, the fast food brands in particular, that made that a best investment starting a number, you know, a few years ago, they've been in a much better position sort of COVID times to deal with like order ahead and pick up at the curb. So it's interesting, because sometimes those investments are not always so obvious as to where things are going. And, you know, Starbucks was even letting you order ahead before COVID. But then all of a sudden, you know, COVID hits, and wow, it's like they had this technology waiting, you know, it was like ready for the moment. And I know they never planned for it like, Oh, there's gonna be a pandemic someday. Let's make sure you can order ahead. But there it was.

Tim Bornholdt 58:37
Yeah, I agree. You never know when your tech investment's going to pay off. And certainly nobody is happy that there's a pandemic going on. But certain restaurants are definitely happier than others at this point with the investment choices that they made early on. Dan, this was such an awesome conversation. I'd love for you to take this time to give my audience any last words of encouragement or feedback, or tell them where they can find you and get Nami going inside of their own apps.

Dan Burcaw 59:07
Yeah, well, words of encouragement is I mean, it's like the best time to be, I mean, we talked about complexity. But at the same, you know, you were talking earlier about GPS in the running app. And, you know, the simple use cases in the old days that were like a lot of work. But now there's just such the toolbox, the AR kit and the machine learning and just, you know, swift UI, and just like, we have all the tools at our disposal to make amazing things. And, I mean, I almost think sometimes the hardest thing is to just choose something to build and then see it through. So I just, you know, kudos to anybody that's listening and wants to build something or is building something and wants to keep pulling on that string and see where it's going to take them because it's just like the best time to be in the technology field or sort of touching technology.

And then in terms of Nami, we are on the web as we should be. And our website is www.NamiML.com. And, you know, so our service is, we work with a lot of big apps, but we also have, you know, tiny little, you know, Mom and Pop apps that have an in app purchase or subscription offering. And they've plugged us in and got a pretty nice kind of free tier. And then a lot of functionality and approach here. And so anyway, anybody can kind of go to our website and sign up for an account, it's like a couple clicks and you're in. In fact, just one more quick thing on that. In the old days, if you would go and build out subscription kind of infrastructure yourself, you know, might be months and months of work to kind of get it right. And we have people that have plugged, taking their existing app and added subscriptions using our platform, and it's taken them like an hour. So it's really not a lot of work. And so hopefully that means that you can focus on kind of your core app experience and not focus on kind of the plumbing related just simply offering a certain monetization approach.

Tim Bornholdt 1:01:24
Right on. So yeah, if you have a desire to put subscriptions into your app, or if you're kind of a Greenfield space, and you want to focus on your core experience, and not worry about all the plumbing of the subscription process, you should definitely check out Nami. Dan, thank you so much for joining me today. I really appreciate it.

Dan Burcaw 1:01:42
Hey, thanks for having me. It was a lot of fun.

Tim Bornholdt 1:01:45
Thanks to Dan for joining me on the podcast today. You can learn more about him and his company at NamiML.com. Show notes for this episode can be found at constantvariables.co. You can get in touch with us by emailing Hello@constantvariables.co. I'm @TimBornholdt on Twitter and the show is @CV_podcast. Today's episode was produced by Jenny Karkowski and edited by the lavish Jordan Daoust.

If you have a minute quick before you leave, we'd love it if you left us a review on the Apple Podcast app. It shouldn't take you much time at all. And it really does help new people find our show, just head to constantvariables.co/review and we'll link you right there. This episode is brought to you by The Jed Mahonis Group. If you're looking for a technical team who can help make sense of mobile software development, give us a shout at jmg.mn.