[00:00:00.860] - Olivier Destrebecq
Welcome to the Subscription League, a podcast by Purchasely. Listen to what's working in subscription apps. In each episode, we invite leaders of the app industry who are mastering the subscription model for mobile apps. To learn more about subscriptions, head to subscriptionleague.com. Let's get started.
[00:00:20.390] - Olivier Destrebecq
Welcome to the show. Today, Jeff and I have the pleasure to interview someone most mobile developers know or at least have to use this service, Felix Krause. He's the founder at ContextSDK. Welcome to the show, Felix.
[00:00:32.130] - Felix Krause
Thank you for having me.
[00:00:33.550] - Olivier Destrebecq
You're welcome. Hey, Jeff, how are you?
[00:00:35.220] - Jeff Grang
I'm great. Super happy to welcome Felix as we've all been as developer, user of his previous ventures. I'm really happy to unveil what he's doing right now at ContextSDK. Welcome, Felix.
[00:00:51.850] - Olivier Destrebecq
Jeff and I know you, Felix, from our background as Mobile Dev, because you've created fastlane, which is an open-source library used to easily ship and deploy mobile apps. It ended up being bought by Twitter and then Google. If I'm not mistaken, you started that as a university project. Can you tell us a little bit about it and that journey from university all the way to full-time paying gig and being famous in our world?
[00:01:19.700] - Felix Krause
Yeah. Fastlane started out as a project to solve a problem that I had myself, which is releasing iOS apps. Back then, tooling and in particular, Xcode was not as good as it is today. The community was already tooling to improve the development process. Stuff like Cocoa Pops or back then there was this tool set called Normand Tools like Shenzhen by Matt Thompson.
[00:01:43.860] - Felix Krause
They were all written in Ruby, which is also why Fasten is written in Ruby. Back then there was the Swift, which is painful now to have to set up Ruby to get it to work, but that's the reason why it started out like this.
[00:01:55.420] - Felix Krause
It all started out again because I solved my own problem, which was uploading localized screenshots, creating those screenshots, dealing with code signing. I remember during my second internship, I was working at Wunderlist, which was like a to-do list app from Berlin.
[00:02:10.580] - Felix Krause
We were three people just sitting for the full day, only uploading screenshots because they supported every single language there is and every single device type. That's how it all started.
[00:02:22.450] - Felix Krause
During university, I took the time to sketch out a tool on what it could look like, build it, open source it, and that's how it started. With every single release and back then it wasn't fast as a whole, but it was a few smaller tools that in the beginning didn't even work together, with every single launch there was more and more traction on Twitter. I built up my Twitter follower base, I built up a newsletter around that and it grew more and more.
[00:02:51.500] - Felix Krause
In the end, I had multiple options to proceed on starting my own business around fastlane or what I decided on is joining Twitter to work on it full-time so I don't have to focus on revenue and finances, but I can just focus on what's best for the user.
[00:03:08.520] - Olivier Destrebecq
Nice. It's nice to be reminding of what iOS development looked like back in the days. We were there, we suffered through the same issues, and it's nice to meet the person that essentially helped us not have that issue anymore. You've gone Twitter first, then Google. Those are two very large US company and I'm sure they have some different cultures. What was the experience of working in those companies like and any big difference that you'd like to highlight?
[00:03:40.070] - Felix Krause
Yeah, so we got moved over from Twitter to Google, mostly because Google bought a whole developer toolset for mobile apps from Twitter. Twitter back then was refocusing on the core product, which was the social network.
[00:03:56.140] - Felix Krause
Very early on, Twitter started this toolset called Fabric, which would allow them to be a player in the mobile dev ecosystem because they don't have an operating system. Because back then they thought they were like Facebook, they are like a platform. I think it didn't pan out like this. That's why they ended up selling Fabric and back then fastlane was part of that.
[00:04:18.160] - Felix Krause
That's how I got moved over to Google and the transition was very smooth and very easy. The reason being is that I remember back then at Twitter, people were always talking about how the cultures are very similar because Twitter's engineering team was mostly previous Googlers.
[00:04:34.270] - Felix Krause
A lot of Googlers switched over to Twitter apparently and they were the ones building out how performance cycle works, how software is being built, how teams are being structured. Also from the developer toolset, back then Google already had Firebase, which was also an acquisition from Google's end, and they had very aligned tools.
[00:04:52.710] - Felix Krause
Whatever tool was better ended up being the one that stayed. Crashlytics is obviously number one. Firebase crash reporter got basically thrown away and replaced with I think it's still called Firebase Crashlytics or just Firebase crash reporting, but under the hood that's Crashlytics.
[00:05:08.590] - Olivier Destrebecq
Interesting. It's very interesting to hear those stories of the culture and how those teams evolved and see how Twitter is in the news every day. Well, not every day today, but has been in the news for various things.
[00:05:21.360] - Jeff Grang
There's also a thing that I'd like to explore is that you've been working in California and you moved back to Europe. So coming from Europe, California, and back to Europe, there's also a big shift in your learnings. Can you tell us a little bit about how different is the European startup ecosystem from when you left Europe and came back?
[00:05:40.190] - Felix Krause
In between, I lived in both California, but then I also lived in New York for two and a half years. To be honest, I think the biggest shift is probably, maybe it sounds boring, but around remote work. Suddenly if you're in Europe, you can just work for a US company, no problem. You just have to consider the time zones. I think that's one of the biggest shifts and it suddenly allows at least individuals to easily work closer together with US companies.
[00:06:04.750] - Olivier Destrebecq
Which is actually interesting too, because I think some companies are coming back a little bit from that remote work. We'll see in a couple of years where that lands if that remains a trend, is to be able to work for this US company more easily or not. Enough talking about the past. You're actually working on ContextSDK now. Can you tell us what it is?
[00:06:22.940] - Felix Krause
Yeah. ContextSDK is a project or an idea that I had for, I believe, a little bit over five years. I had this idea initially when I was driving from my parents' house and driving back home, I think, and I wanted to play some music and the music app I was using back then, I couldn't press the button to play my playlist and there was no car play back then.
[00:06:46.910] - Felix Krause
Because there was a pop-up covering the areas, I pulled over to see what it said and it was some very unnecessary pop-up. It was something like, "Welcome to the new version of this app, see what's new in this update."
[00:06:58.170] - Felix Krause
In this case, I was actually interested in what's new. I was like, "Hey, you're showing a pop-up, it must be something cool. But right now it was not the right moment. I had to remember that once I arrive home, I want to open up that app to see what's new.
[00:07:10.030] - Felix Krause
I felt like, "Well, this should be so obvious for an app to realize what's the context like is now a good moment to show something." Because we have all these really powerful sensors and you can do on-device processing. You don't even need to do anything with remote service and stuff. You could just check like, "Hey, is this now? Are you in a car? Are you walking around? Or are you just sitting or being in a more calm environment?"
[00:07:33.550] - Felix Krause
Back then I was still employed working on a few other projects. Once I finished my last project, I was looking into what's next and I realized, "Wait, this is still something on my list and nobody seems to have built it." There might be a reason why no one built it, you never know.
[00:07:50.240] - Felix Krause
First in I had something very similar. I remember when I built this tool for push notifications and you had to copy and paste commands into the terminal, I always felt like, "Wait, this doesn't seem right. This is something very basic I'm doing. Why do I need to do some weird encryption stuff or a private public key for setting up push."
[00:08:06.690] - Felix Krause
I just built the tool to see what people are saying. As in instead of asking for advice, just post an answer and get corrected by the community. With ContextSDK as a case, it's similar. Maybe there's a reason why no one build it, but let's just build and see what it's like.
[00:08:22.270] - Felix Krause
In this case, initially, it was very much focused on just detecting the type of activity that you are on. It would be something like, "Are you currently driving? Are you currently walking? Are you currently sitting?" Very basic high level and this would be all through on-device machine learning execution. Then you could do something like, "Hey, don't show this pop-up in a car."
[00:08:43.550] - Felix Krause
That was the initial idea. Then very early on, I realized when I was working with potential users of this SDK, "Well, they still have to figure out how the app behaves in those different situations. You would just end up with a lot of if-else statements on like, is the user walking right now? Is the user driving right now? Very soon, ContextSDK has shifted over to something a little bit more abstract, which is just telling ContextSDK optimize for this specific outcome.
[00:09:10.790] - Felix Krause
This could be something like, "Hey, you showed this update prompt, like hey, what's new in this version?" The positive outcome would be the user opening up that prompt and reading what's new. The negative outcome would be the user dismissing it. Based on that, we can learn to see which situations are positive versus negative.
[00:09:28.510] - Jeff Grang
Speaking about outcomes, I know that the company and the tech is still very young, but what are the use cases you see for ContextSDK, like concrete use cases? Do you have any numbers or case study that you could share with us?
[00:09:41.990] - Felix Krause
Yeah. When it comes to use cases, the first one we focused on was the topic of in-app subscription and in-app sales, which we believe is the most obvious way to build a business.
[00:09:53.510] - Felix Krause
As in, it's hard to sell something that's only nice to have and build a business around it. Selling the increased conversion rate for upsells was something that's more compelling for potential customers. That was the first one. The second one we focused on the last few weeks and we just launched a little product around is the app tracking transparency prompt.
[00:10:12.130] - Felix Krause
A lot of ad-based apps, a lot of ones in the purchases, but the other approach would be advertising. They have to use this ATT opt-in and depending on the opt-in rate, they make different kinds of money.
[00:10:22.450] - Felix Krause
One customer we talked with, they said if we can help them increase the ATT opt-in rate by 10%, they would make around 5% to 7% more revenue. That was something like, "Well, this is interesting. This is something we can help with, find the right moment to ask the user for consent."
[00:10:37.090] - Olivier Destrebecq
If you can just make it so that it doesn't launch, just show at the same time as the app launch that would be great.
[00:10:43.660] - Felix Krause
Exactly. Right now, the approach for many companies, also for in-app subscription, I think Jeff has some experience there, is to just show it right away because that's where you have the highest exposure. Because everybody who launches that once will get it.
[00:10:55.820] - Felix Krause
What we are trying to focus on is the long-term value, the LTV of those apps and those users. By showing the prompts later and being less annoying to the user, we believe that you'll end up with less annoyed users, more positive attitude towards your app, less churn and overall, hopefully more revenue. Those are the first two use cases.
[00:11:16.510] - Felix Krause
We do have many other use cases, so it's just in general, commission prompts, similarly annoying. We can help time those at right moments. We have all these in-app messaging. That would be something like the update message. That would be anything you want to communicate, which may also include new features.
[00:11:34.090] - Felix Krause
I think Apple has launched a framework to help you introduce new features with this pop-over, but they don't help you again find the timing. I think Apple's app are actually really bad offenders of that because they give you launch.
[00:11:44.320] - Felix Krause
Every time I launch pages or keynote on my Mac, there's always this big, full screen like, "Hey, you see what's new? I really don't care. I just want to open up this Excel file."
[00:11:52.680] - Jeff Grang
Same thing when you have a new iPhone that you install, you have all the apps telling you, "Hey, what's new? What's new? What's new? I don't care about what's new. I just want to use my phone right now." Same trouble that you have.
[00:12:04.120] - Felix Krause
Exactly. For the case studies, because you are so obviously context is still pretty new. We don't really have long-term studies yet, but we're working on that because that's obviously critical. We do have the first numbers and reports that came out.
[00:12:18.160] - Felix Krause
One interesting thing is that for one really large app, we were able to improve the conversion rate by a few percent. I think it was like 8 or 9% or something, improved the conversion rate. By that, we were not too excited because for smaller apps, we sometimes saw an even better conversion rate increase.
[00:12:34.410] - Felix Krause
We were a little bit disappointed about like, "Okay, let's look into this and get better at this." What's interesting is that this customer was actually really excited because what they did is they saw the lifetime value increase by a lot more. It was like a 20% lifetime value increase in those cases for these buckets because they did a regular EV test.
[00:12:52.180] - Felix Krause
That proves our assumption, and something that I believe most humans would just agree with, which is like, "Hey, if you're less annoying and you show less pop-ups and ask at the right time, long-term people are going to be less annoyed, they're going to churn less, and they are maybe more likely to upgrade."
[00:13:07.710] - Felix Krause
At least that's our core assumption here. The problem obviously with selling or providing a service like this is that customers want to see you're the one that will tell them all the data. Like, "Hey, is this going to make more money or is this just going to be something to be nice to the user?" So far it looks really promising. Obviously, it still takes time to put all these things together and analyze the data more, but that's where we are.
[00:13:26.900] - Jeff Grang
Yeah, that's a great promise here. You're moving from a tech tool, providing a tech tool to real marketing tool. I believe that it could go even further being a product tool because once you will educate all the markets about the importance of context in building apps and promoting experiences, it will go way further or it could go way further than just the monetisation and the ATT prompts, but also shape differently how we build UIs. That's a long question. What are the differences between providing a tech tool for developers and providing a marketing tool for marketers or growth managers?
[00:14:03.700] - Felix Krause
Yeah. To do the first part of the question, it's a very interesting perspective and that's how we see it also, which is, "Hey, is this the future of how you build apps? Instead of being very static and just having those hard-coded rules be powered by ML."
[00:14:17.130] - Felix Krause
All these tech companies are obviously always talking about it, but those kinds of things work. Just even basic statistics and data analysis already helps. Maybe ContextSDK can be the foundation of how you build new apps. We don't yet have specific plans for it because, as you know, just throwing screens into a bucket and letting context as the key figure out the order, that's something we could do and that's something we're thinking about. Obviously it comes with complexities.
[00:14:41.550] - Felix Krause
For example, to edit an item in an app, you need to first have that item open. It's just going to be very, very difficult on how can you structure the app but still have this fluid movement of ContextSDK. It could be something way bigger and we're building it in a way that allows us to do those kinds of things. It's mostly about talking with potential customers and users on how it would integrate with their stack.
[00:15:04.940] - Felix Krause
To your second question around building a developer tool versus marketing growth revenue tool, that is a big one because obviously, I am a developer myself, so I can build developer tools really well, I would say fastlane was one of them. It was very much focused on the iOS developer, just seeing it and then installing it, setting up and using it. It's very easy for me to also speak their language.
[00:15:28.330] - Felix Krause
Now it's a big shift. ContextSDK initially was also a developer tool, as mentioned, it's here to help you detect if the user is walking, sitting, stuff like that, and now shift it to be focused on marketing growth and so on. That's also why the name is ContextSDK, but I would say SDK is a very developer-heavy phrase.
[00:15:48.780] - Felix Krause
Unfortunately, all the just context domains are taking because that would be an ideal name, I think. Now it has to shift. It's interesting, right? Just even the language and the way things are phrased is so different that I sometimes really have to just trust, in this case, my co-founder, who is way better at speaking to people in marketing and growth and so on.
[00:16:11.490] - Felix Krause
I would do it differently, but also I'm a developer. I'm not target group for this service anymore. It comes with challenges for sure and it makes it harder because reaching developers for me is easy because you have Twitter, you have your friends there, you know the newsletters, you know the in and outs, you know the conferences. Then suddenly it's like a whole different, how do you say, like job group and I don't know where they hang up.
[00:16:34.210] - Olivier Destrebecq
Yeah, it's a different ecosystem to hang out with the marketing folks than with the developer folks. You mentioned that ContextSDK is essentially using machine learning to give the recommendation as to when to do certain tasks. What does this setup look like in an app to get ContextSDK? Like is there a training that needs to happen or what do we have to do?
[00:16:54.770] - Felix Krause
One thing ContextSDK from the very beginning was focused on privacy. One grand rule we have is that we never create, use or process any PII, so any information that's specific to a person. Therefore we are exempt from most privacy laws because if data entry is not associated with a person or a device and you cannot find out who it was, then it doesn't fall into those categories.
[00:17:18.410] - Felix Krause
That was like a ground rule. As you mentioned, we do use machine learning. Machine learning needs a high amount of data and there are different types. One type that Apple does a lot is the on-device machine learning, which actually learns based on your behavior on your phone.
[00:17:32.280] - Felix Krause
What we need to do is people don't use a specific app too often. They would just launch the app a few times. You already need a model to help you optimize. What we do when you integrate ContextSDK, you set it up in a calibration mode, which would allow us to calibrate all the signals. We have around 170 device signals that we consider, anything from motion data, time of the day, battery status, all harmless but very useful stuff to calculate the likelihood of somebody engaging or using certain features at that moment.
[00:18:02.140] - Felix Krause
We would calibrate those signals and they get sent to our servers again fully anonymized together with the outcome. Did they use upgrade now or did they not upgrade? Based on that, we can train the machine learning model. As you know, training a machine learning model can be very expensive and very CPU intensive, so you need a lot of computer power and that allows us to pre-train those models.
[00:18:23.900] - Felix Krause
That's what's interesting because then you have this pre-trained model that you can ship with your app and then all the decisions can happen on-device instantly. We would actually provide their own SDK binary for every single customer we have with that pre-trained model already included.
[00:18:40.430] - Felix Krause
I built a mini fastlane for SDKs basically. I think that's also very unique because usually you have one SDK for all your customers and then you have options. We actually have custom builds for every single one. Then they ship that update and they have the model included and from then on, all the decisions happen on-device.
[00:18:57.690] - Olivier Destrebecq
Nice. You mentioned the privacy and you're not processing any PIIs or anything identifiable. Have you received any concerns from user about the SDK knowing more than what they feel comfortable sharing potentially? I know it's not identifiable information, but knowing that somebody or something knows about the context of you using the app. Have you got any pushback on that?
[00:19:22.650] - Felix Krause
Currently not yet. Our customers would be the companies that use our SDK. I assume you're talking about the end users, right?
[00:19:29.370] - Olivier Destrebecq
Yeah, I'm sure maybe the companies or the users have talked to the publisher of the app or any feedback you might have had, I guess.
[00:19:37.860] - Felix Krause
Yeah. So far there hasn't been any pushback neither by European nor by US companies. I think that's for a few reasons. One is that we are always very clear on our marketing materials, but also our documentation on how we process data. We are very transparent on what data we use. Again, that it's not specific to a person.
[00:19:58.100] - Felix Krause
The second one that comes from direct customers would be the trust that we have in the community. As you may know, I've worked on a lot of different privacy and security projects on iOS and disclosed a few loopholes and Apple ecosystems. I think there is a good amount of trust there. It's something we'll constantly have to work on, though, around communication and being transparent and so on.
[00:20:18.260] - Felix Krause
Once we have enough customers, what we aim to do also is to have a version of the SDK that's not self-learning or not calibrating. For example, if you have a certain app, we can already offer you a trained machine learning model for your type of app, for example.
[00:20:36.330] - Felix Krause
Because what we see is we have those custom models for each of our customers, and that's so important because each app is used in different situations. Like, for example, if you launch a fitness app, it's going to be used in a very different situation than a dating app, for example.
[00:20:49.190] - Felix Krause
Not just when you launch it, but also your customer base, your user base is going to be different. You launch at different times of the day, your different age group, stuff like that.
[00:20:59.360] - Felix Krause
One random interesting find was the battery level. It's a very simple signal that usually doesn't mean too much. For one fitness app, we found that it decreased the conversion by a lot. I think it was 60, 70% or something. If the battery level is below, I think it was like 5% or something, but it was like a fitness app.
[00:21:19.330] - Felix Krause
You record your workout or something, you don't want your phone to die while you do that. Then we have another app that's more focused towards Gen Z, very young users, and turns out they really couldn't care less about battery. It's literally completely flat. That's why we have this calibration phase to calibrate our signals, and that's how we achieve the best performance.
[00:21:41.840] - Jeff Grang
Great. There's one thing I'd like to talk about, and I'm sure, I'm sure you've been facing the same issues. When you integrate an SDK, an external SDK, there's marketing team to say, "Hey, I want you to insert that new ad SDK within your app." You uncover the things and say, "Oh, wow, they're using this and these APIs and the SDK is really bad." It's always a fight between marketing and tech, and we always lose as tech people.
[00:22:10.550] - Jeff Grang
I'm sure every developer listening to the podcast right now have faced this situation. I know that when we build it, the first version of Purchasely, we really wanted to develop an SDK that we would love using and integrating, so easy to integrate, well-documented, lightweight, respecting the privacy of the users, et cetera. I'd like to hear your thoughts on this, Felix, because I'm sure that you've pushed that really far with ContextSDK being a good citizen.
[00:22:38.780] - Felix Krause
Yeah. As you mentioned, I think you and I have a very similar view on this and opinion on this. ContextSDK also is 100% written natively. It uses like, I think it's 700 kilobytes added to your binary size, which is really nothing compared to what you describe around some advertising SDKs using, I think, up to 100 megabytes or something.
[00:23:01.400] - Felix Krause
Very little CPU usage when accessing it. I think it's like 0.2% or something, like battery usage and CPU usage, bringing that down to basically nothing was a very satisfying and fun challenge, I would say.
[00:23:15.700] - Felix Krause
I think it's so important because if you provide an SDK, then it's going to be used on many tens of millions and more devices suddenly. It has huge impact. I think you always want to be a good citizen for multiple reasons. One is towards Apple, which is obviously the platform, but also towards users. You're just using up the battery or whatever you're doing. That's very much a big focus for us.
[00:23:39.460] - Olivier Destrebecq
If our listeners want to learn more about ContextSDK and how to use it and what it can do for them, where can they go?
[00:23:46.650] - Felix Krause
We have contextsdk.com. Right now we are not self-serviced, so it has to be going through email right now to onboard you because also we have for every app, there is going to be some different integration.
[00:24:00.990] - Felix Krause
For example, if you have paywalls and you have a paywall on the first app start, it's going to be a different integration than if you're just upselling something sometimes. We help you get that on board. Eventually, obviously it should be fully automated. Right now, automating something where it still needs some hand-holding and help from us and some guidance from us wouldn't make sense.
[00:24:19.580] - Felix Krause
Yeah, contextsdk.com, and then we have a contact form. We just fill out basically number of active users and the name of your app. The reason why we're asking for the number of active users is because our models are going to work differently. The more users you have, the faster you're going to have your own custom model and the more precise that model is going to be.
[00:24:37.410] - Olivier Destrebecq
Nice. If people want to learn more about you, I actually found out that you're very active on social network, but you also have your own website with a bunch of stuff about you. Where can people learn more about you, Felix?
[00:24:51.400] - Felix Krause
I have a few starting points. One is krausefx.com, where I have my blog and some information about me, as well as my privacy publications on the iOS platform. Then I have this other website that I haven't worked on recently, but it's a big website which is called How is Felix Start today? It's a huge quantified self-dashboard.
[00:25:12.940] - Felix Krause
It contains all this information about anything from the number of steps I took to my energy level, happiness level or how many hours I was in meetings. It's How is Felix Start Today? It started out as a fun project to just show where I am because initially it was called Where is Felix Start today? I was just showing my flight schedule for conferences and stuff that I spoke at.
[00:25:33.580] - Felix Krause
Then it just escalated up to be a full quantified self-dashboard, which was always like a hobby of mine. I'm still recording all of these things every day now, but I haven't updated the graphs yet. The graphs are always snapshots because I still want to make sure that data that's included there is not too personal.
[00:25:49.100] - Olivier Destrebecq
It's actually interesting that you mentioned the website because I went and read it and the conclusion there was it was a great experiment. I learned quite a bit about myself, but it was a lot of effort to do it. It's very interesting to hear that you are actually still doing it and putting the effort in, so it's great.
[00:26:04.120] - Felix Krause
Yeah. I think nowadays I would do a different day. Apple has launched some really cool Apple Health stuff and even the Journal app, which I haven't had a chance to try yet. I remember that Apple Health app now has the mood tracking building, for example, with a very nice use interface and everything. Maybe I'll build something on top of that instead if I were to build it now.
[00:26:23.320] - Felix Krause
The main conclusion was that you mentioned now was around the effort around building the data analysis tools because you need to set it up that you can even build those graphs.
[00:26:34.770] - Felix Krause
To be honest, nowadays I would have way better tools and experience thanks to ContextSDK because it's the topic of data science. Back then, if you don't know the tools, like if you as an iOS developer don't know how to use Xcode and code signing, you're going to have a hard time. That was me back then not having any data science experience back then.
[00:26:52.760] - Olivier Destrebecq
Nice. Well, thanks for joining us today. It was really nice to get all your answer and your perspective on your experiences and see where you're heading with ContextSDK. Thanks for joining us today, Felix.
[00:27:02.570] - Jeff Grang
Thank you, Felix. I hope we're going to have the opportunity to test both our technologies on the customer and see how both context and context trigger and contextual paywalls and experiences can leverage a better growth and our LTV in subscription area.
[00:27:19.300] - Felix Krause
Perfect. Thank you for having me.
[00:27:20.670] - Jeff Grang
Yeah, of course.
[00:27:21.920] - Olivier Destrebecq
Thanks, Chris.
[00:27:23.050] - Olivier Destrebecq
On behalf of the Purchasely team, thank you for listening to the Subscription League podcast. If you've enjoyed what you heard, leave us a five-star review on iTunes or other audio platform. To find out more about Purchasely and how we can improve your subscription business, visit purchasely.com.
[00:27:39.900] - Olivier Destrebecq
Please hit subscribe in your podcast player and don't miss any future episodes. You can also listen to previous episodes at subscriptionleague. com. See you soon.
Welcome to the Subscription League, a podcast by Purchasely. Listen to what's working in subscription apps. In each episode, we invite leaders of the app industry who are mastering the subscription model for mobile apps. To learn more about subscriptions, head to subscriptionleague.com. Let's get started.
[00:00:20.390] - Olivier Destrebecq
Welcome to the show. Today, Jeff and I have the pleasure to interview someone most mobile developers know or at least have to use this service, Felix Krause. He's the founder at ContextSDK. Welcome to the show, Felix.
[00:00:32.130] - Felix Krause
Thank you for having me.
[00:00:33.550] - Olivier Destrebecq
You're welcome. Hey, Jeff, how are you?
[00:00:35.220] - Jeff Grang
I'm great. Super happy to welcome Felix as we've all been as developer, user of his previous ventures. I'm really happy to unveil what he's doing right now at ContextSDK. Welcome, Felix.
[00:00:51.850] - Olivier Destrebecq
Jeff and I know you, Felix, from our background as Mobile Dev, because you've created fastlane, which is an open-source library used to easily ship and deploy mobile apps. It ended up being bought by Twitter and then Google. If I'm not mistaken, you started that as a university project. Can you tell us a little bit about it and that journey from university all the way to full-time paying gig and being famous in our world?
[00:01:19.700] - Felix Krause
Yeah. Fastlane started out as a project to solve a problem that I had myself, which is releasing iOS apps. Back then, tooling and in particular, Xcode was not as good as it is today. The community was already tooling to improve the development process. Stuff like Cocoa Pops or back then there was this tool set called Normand Tools like Shenzhen by Matt Thompson.
[00:01:43.860] - Felix Krause
They were all written in Ruby, which is also why Fasten is written in Ruby. Back then there was the Swift, which is painful now to have to set up Ruby to get it to work, but that's the reason why it started out like this.
[00:01:55.420] - Felix Krause
It all started out again because I solved my own problem, which was uploading localized screenshots, creating those screenshots, dealing with code signing. I remember during my second internship, I was working at Wunderlist, which was like a to-do list app from Berlin.
[00:02:10.580] - Felix Krause
We were three people just sitting for the full day, only uploading screenshots because they supported every single language there is and every single device type. That's how it all started.
[00:02:22.450] - Felix Krause
During university, I took the time to sketch out a tool on what it could look like, build it, open source it, and that's how it started. With every single release and back then it wasn't fast as a whole, but it was a few smaller tools that in the beginning didn't even work together, with every single launch there was more and more traction on Twitter. I built up my Twitter follower base, I built up a newsletter around that and it grew more and more.
[00:02:51.500] - Felix Krause
In the end, I had multiple options to proceed on starting my own business around fastlane or what I decided on is joining Twitter to work on it full-time so I don't have to focus on revenue and finances, but I can just focus on what's best for the user.
[00:03:08.520] - Olivier Destrebecq
Nice. It's nice to be reminding of what iOS development looked like back in the days. We were there, we suffered through the same issues, and it's nice to meet the person that essentially helped us not have that issue anymore. You've gone Twitter first, then Google. Those are two very large US company and I'm sure they have some different cultures. What was the experience of working in those companies like and any big difference that you'd like to highlight?
[00:03:40.070] - Felix Krause
Yeah, so we got moved over from Twitter to Google, mostly because Google bought a whole developer toolset for mobile apps from Twitter. Twitter back then was refocusing on the core product, which was the social network.
[00:03:56.140] - Felix Krause
Very early on, Twitter started this toolset called Fabric, which would allow them to be a player in the mobile dev ecosystem because they don't have an operating system. Because back then they thought they were like Facebook, they are like a platform. I think it didn't pan out like this. That's why they ended up selling Fabric and back then fastlane was part of that.
[00:04:18.160] - Felix Krause
That's how I got moved over to Google and the transition was very smooth and very easy. The reason being is that I remember back then at Twitter, people were always talking about how the cultures are very similar because Twitter's engineering team was mostly previous Googlers.
[00:04:34.270] - Felix Krause
A lot of Googlers switched over to Twitter apparently and they were the ones building out how performance cycle works, how software is being built, how teams are being structured. Also from the developer toolset, back then Google already had Firebase, which was also an acquisition from Google's end, and they had very aligned tools.
[00:04:52.710] - Felix Krause
Whatever tool was better ended up being the one that stayed. Crashlytics is obviously number one. Firebase crash reporter got basically thrown away and replaced with I think it's still called Firebase Crashlytics or just Firebase crash reporting, but under the hood that's Crashlytics.
[00:05:08.590] - Olivier Destrebecq
Interesting. It's very interesting to hear those stories of the culture and how those teams evolved and see how Twitter is in the news every day. Well, not every day today, but has been in the news for various things.
[00:05:21.360] - Jeff Grang
There's also a thing that I'd like to explore is that you've been working in California and you moved back to Europe. So coming from Europe, California, and back to Europe, there's also a big shift in your learnings. Can you tell us a little bit about how different is the European startup ecosystem from when you left Europe and came back?
[00:05:40.190] - Felix Krause
In between, I lived in both California, but then I also lived in New York for two and a half years. To be honest, I think the biggest shift is probably, maybe it sounds boring, but around remote work. Suddenly if you're in Europe, you can just work for a US company, no problem. You just have to consider the time zones. I think that's one of the biggest shifts and it suddenly allows at least individuals to easily work closer together with US companies.
[00:06:04.750] - Olivier Destrebecq
Which is actually interesting too, because I think some companies are coming back a little bit from that remote work. We'll see in a couple of years where that lands if that remains a trend, is to be able to work for this US company more easily or not. Enough talking about the past. You're actually working on ContextSDK now. Can you tell us what it is?
[00:06:22.940] - Felix Krause
Yeah. ContextSDK is a project or an idea that I had for, I believe, a little bit over five years. I had this idea initially when I was driving from my parents' house and driving back home, I think, and I wanted to play some music and the music app I was using back then, I couldn't press the button to play my playlist and there was no car play back then.
[00:06:46.910] - Felix Krause
Because there was a pop-up covering the areas, I pulled over to see what it said and it was some very unnecessary pop-up. It was something like, "Welcome to the new version of this app, see what's new in this update."
[00:06:58.170] - Felix Krause
In this case, I was actually interested in what's new. I was like, "Hey, you're showing a pop-up, it must be something cool. But right now it was not the right moment. I had to remember that once I arrive home, I want to open up that app to see what's new.
[00:07:10.030] - Felix Krause
I felt like, "Well, this should be so obvious for an app to realize what's the context like is now a good moment to show something." Because we have all these really powerful sensors and you can do on-device processing. You don't even need to do anything with remote service and stuff. You could just check like, "Hey, is this now? Are you in a car? Are you walking around? Or are you just sitting or being in a more calm environment?"
[00:07:33.550] - Felix Krause
Back then I was still employed working on a few other projects. Once I finished my last project, I was looking into what's next and I realized, "Wait, this is still something on my list and nobody seems to have built it." There might be a reason why no one built it, you never know.
[00:07:50.240] - Felix Krause
First in I had something very similar. I remember when I built this tool for push notifications and you had to copy and paste commands into the terminal, I always felt like, "Wait, this doesn't seem right. This is something very basic I'm doing. Why do I need to do some weird encryption stuff or a private public key for setting up push."
[00:08:06.690] - Felix Krause
I just built the tool to see what people are saying. As in instead of asking for advice, just post an answer and get corrected by the community. With ContextSDK as a case, it's similar. Maybe there's a reason why no one build it, but let's just build and see what it's like.
[00:08:22.270] - Felix Krause
In this case, initially, it was very much focused on just detecting the type of activity that you are on. It would be something like, "Are you currently driving? Are you currently walking? Are you currently sitting?" Very basic high level and this would be all through on-device machine learning execution. Then you could do something like, "Hey, don't show this pop-up in a car."
[00:08:43.550] - Felix Krause
That was the initial idea. Then very early on, I realized when I was working with potential users of this SDK, "Well, they still have to figure out how the app behaves in those different situations. You would just end up with a lot of if-else statements on like, is the user walking right now? Is the user driving right now? Very soon, ContextSDK has shifted over to something a little bit more abstract, which is just telling ContextSDK optimize for this specific outcome.
[00:09:10.790] - Felix Krause
This could be something like, "Hey, you showed this update prompt, like hey, what's new in this version?" The positive outcome would be the user opening up that prompt and reading what's new. The negative outcome would be the user dismissing it. Based on that, we can learn to see which situations are positive versus negative.
[00:09:28.510] - Jeff Grang
Speaking about outcomes, I know that the company and the tech is still very young, but what are the use cases you see for ContextSDK, like concrete use cases? Do you have any numbers or case study that you could share with us?
[00:09:41.990] - Felix Krause
Yeah. When it comes to use cases, the first one we focused on was the topic of in-app subscription and in-app sales, which we believe is the most obvious way to build a business.
[00:09:53.510] - Felix Krause
As in, it's hard to sell something that's only nice to have and build a business around it. Selling the increased conversion rate for upsells was something that's more compelling for potential customers. That was the first one. The second one we focused on the last few weeks and we just launched a little product around is the app tracking transparency prompt.
[00:10:12.130] - Felix Krause
A lot of ad-based apps, a lot of ones in the purchases, but the other approach would be advertising. They have to use this ATT opt-in and depending on the opt-in rate, they make different kinds of money.
[00:10:22.450] - Felix Krause
One customer we talked with, they said if we can help them increase the ATT opt-in rate by 10%, they would make around 5% to 7% more revenue. That was something like, "Well, this is interesting. This is something we can help with, find the right moment to ask the user for consent."
[00:10:37.090] - Olivier Destrebecq
If you can just make it so that it doesn't launch, just show at the same time as the app launch that would be great.
[00:10:43.660] - Felix Krause
Exactly. Right now, the approach for many companies, also for in-app subscription, I think Jeff has some experience there, is to just show it right away because that's where you have the highest exposure. Because everybody who launches that once will get it.
[00:10:55.820] - Felix Krause
What we are trying to focus on is the long-term value, the LTV of those apps and those users. By showing the prompts later and being less annoying to the user, we believe that you'll end up with less annoyed users, more positive attitude towards your app, less churn and overall, hopefully more revenue. Those are the first two use cases.
[00:11:16.510] - Felix Krause
We do have many other use cases, so it's just in general, commission prompts, similarly annoying. We can help time those at right moments. We have all these in-app messaging. That would be something like the update message. That would be anything you want to communicate, which may also include new features.
[00:11:34.090] - Felix Krause
I think Apple has launched a framework to help you introduce new features with this pop-over, but they don't help you again find the timing. I think Apple's app are actually really bad offenders of that because they give you launch.
[00:11:44.320] - Felix Krause
Every time I launch pages or keynote on my Mac, there's always this big, full screen like, "Hey, you see what's new? I really don't care. I just want to open up this Excel file."
[00:11:52.680] - Jeff Grang
Same thing when you have a new iPhone that you install, you have all the apps telling you, "Hey, what's new? What's new? What's new? I don't care about what's new. I just want to use my phone right now." Same trouble that you have.
[00:12:04.120] - Felix Krause
Exactly. For the case studies, because you are so obviously context is still pretty new. We don't really have long-term studies yet, but we're working on that because that's obviously critical. We do have the first numbers and reports that came out.
[00:12:18.160] - Felix Krause
One interesting thing is that for one really large app, we were able to improve the conversion rate by a few percent. I think it was like 8 or 9% or something, improved the conversion rate. By that, we were not too excited because for smaller apps, we sometimes saw an even better conversion rate increase.
[00:12:34.410] - Felix Krause
We were a little bit disappointed about like, "Okay, let's look into this and get better at this." What's interesting is that this customer was actually really excited because what they did is they saw the lifetime value increase by a lot more. It was like a 20% lifetime value increase in those cases for these buckets because they did a regular EV test.
[00:12:52.180] - Felix Krause
That proves our assumption, and something that I believe most humans would just agree with, which is like, "Hey, if you're less annoying and you show less pop-ups and ask at the right time, long-term people are going to be less annoyed, they're going to churn less, and they are maybe more likely to upgrade."
[00:13:07.710] - Felix Krause
At least that's our core assumption here. The problem obviously with selling or providing a service like this is that customers want to see you're the one that will tell them all the data. Like, "Hey, is this going to make more money or is this just going to be something to be nice to the user?" So far it looks really promising. Obviously, it still takes time to put all these things together and analyze the data more, but that's where we are.
[00:13:26.900] - Jeff Grang
Yeah, that's a great promise here. You're moving from a tech tool, providing a tech tool to real marketing tool. I believe that it could go even further being a product tool because once you will educate all the markets about the importance of context in building apps and promoting experiences, it will go way further or it could go way further than just the monetisation and the ATT prompts, but also shape differently how we build UIs. That's a long question. What are the differences between providing a tech tool for developers and providing a marketing tool for marketers or growth managers?
[00:14:03.700] - Felix Krause
Yeah. To do the first part of the question, it's a very interesting perspective and that's how we see it also, which is, "Hey, is this the future of how you build apps? Instead of being very static and just having those hard-coded rules be powered by ML."
[00:14:17.130] - Felix Krause
All these tech companies are obviously always talking about it, but those kinds of things work. Just even basic statistics and data analysis already helps. Maybe ContextSDK can be the foundation of how you build new apps. We don't yet have specific plans for it because, as you know, just throwing screens into a bucket and letting context as the key figure out the order, that's something we could do and that's something we're thinking about. Obviously it comes with complexities.
[00:14:41.550] - Felix Krause
For example, to edit an item in an app, you need to first have that item open. It's just going to be very, very difficult on how can you structure the app but still have this fluid movement of ContextSDK. It could be something way bigger and we're building it in a way that allows us to do those kinds of things. It's mostly about talking with potential customers and users on how it would integrate with their stack.
[00:15:04.940] - Felix Krause
To your second question around building a developer tool versus marketing growth revenue tool, that is a big one because obviously, I am a developer myself, so I can build developer tools really well, I would say fastlane was one of them. It was very much focused on the iOS developer, just seeing it and then installing it, setting up and using it. It's very easy for me to also speak their language.
[00:15:28.330] - Felix Krause
Now it's a big shift. ContextSDK initially was also a developer tool, as mentioned, it's here to help you detect if the user is walking, sitting, stuff like that, and now shift it to be focused on marketing growth and so on. That's also why the name is ContextSDK, but I would say SDK is a very developer-heavy phrase.
[00:15:48.780] - Felix Krause
Unfortunately, all the just context domains are taking because that would be an ideal name, I think. Now it has to shift. It's interesting, right? Just even the language and the way things are phrased is so different that I sometimes really have to just trust, in this case, my co-founder, who is way better at speaking to people in marketing and growth and so on.
[00:16:11.490] - Felix Krause
I would do it differently, but also I'm a developer. I'm not target group for this service anymore. It comes with challenges for sure and it makes it harder because reaching developers for me is easy because you have Twitter, you have your friends there, you know the newsletters, you know the in and outs, you know the conferences. Then suddenly it's like a whole different, how do you say, like job group and I don't know where they hang up.
[00:16:34.210] - Olivier Destrebecq
Yeah, it's a different ecosystem to hang out with the marketing folks than with the developer folks. You mentioned that ContextSDK is essentially using machine learning to give the recommendation as to when to do certain tasks. What does this setup look like in an app to get ContextSDK? Like is there a training that needs to happen or what do we have to do?
[00:16:54.770] - Felix Krause
One thing ContextSDK from the very beginning was focused on privacy. One grand rule we have is that we never create, use or process any PII, so any information that's specific to a person. Therefore we are exempt from most privacy laws because if data entry is not associated with a person or a device and you cannot find out who it was, then it doesn't fall into those categories.
[00:17:18.410] - Felix Krause
That was like a ground rule. As you mentioned, we do use machine learning. Machine learning needs a high amount of data and there are different types. One type that Apple does a lot is the on-device machine learning, which actually learns based on your behavior on your phone.
[00:17:32.280] - Felix Krause
What we need to do is people don't use a specific app too often. They would just launch the app a few times. You already need a model to help you optimize. What we do when you integrate ContextSDK, you set it up in a calibration mode, which would allow us to calibrate all the signals. We have around 170 device signals that we consider, anything from motion data, time of the day, battery status, all harmless but very useful stuff to calculate the likelihood of somebody engaging or using certain features at that moment.
[00:18:02.140] - Felix Krause
We would calibrate those signals and they get sent to our servers again fully anonymized together with the outcome. Did they use upgrade now or did they not upgrade? Based on that, we can train the machine learning model. As you know, training a machine learning model can be very expensive and very CPU intensive, so you need a lot of computer power and that allows us to pre-train those models.
[00:18:23.900] - Felix Krause
That's what's interesting because then you have this pre-trained model that you can ship with your app and then all the decisions can happen on-device instantly. We would actually provide their own SDK binary for every single customer we have with that pre-trained model already included.
[00:18:40.430] - Felix Krause
I built a mini fastlane for SDKs basically. I think that's also very unique because usually you have one SDK for all your customers and then you have options. We actually have custom builds for every single one. Then they ship that update and they have the model included and from then on, all the decisions happen on-device.
[00:18:57.690] - Olivier Destrebecq
Nice. You mentioned the privacy and you're not processing any PIIs or anything identifiable. Have you received any concerns from user about the SDK knowing more than what they feel comfortable sharing potentially? I know it's not identifiable information, but knowing that somebody or something knows about the context of you using the app. Have you got any pushback on that?
[00:19:22.650] - Felix Krause
Currently not yet. Our customers would be the companies that use our SDK. I assume you're talking about the end users, right?
[00:19:29.370] - Olivier Destrebecq
Yeah, I'm sure maybe the companies or the users have talked to the publisher of the app or any feedback you might have had, I guess.
[00:19:37.860] - Felix Krause
Yeah. So far there hasn't been any pushback neither by European nor by US companies. I think that's for a few reasons. One is that we are always very clear on our marketing materials, but also our documentation on how we process data. We are very transparent on what data we use. Again, that it's not specific to a person.
[00:19:58.100] - Felix Krause
The second one that comes from direct customers would be the trust that we have in the community. As you may know, I've worked on a lot of different privacy and security projects on iOS and disclosed a few loopholes and Apple ecosystems. I think there is a good amount of trust there. It's something we'll constantly have to work on, though, around communication and being transparent and so on.
[00:20:18.260] - Felix Krause
Once we have enough customers, what we aim to do also is to have a version of the SDK that's not self-learning or not calibrating. For example, if you have a certain app, we can already offer you a trained machine learning model for your type of app, for example.
[00:20:36.330] - Felix Krause
Because what we see is we have those custom models for each of our customers, and that's so important because each app is used in different situations. Like, for example, if you launch a fitness app, it's going to be used in a very different situation than a dating app, for example.
[00:20:49.190] - Felix Krause
Not just when you launch it, but also your customer base, your user base is going to be different. You launch at different times of the day, your different age group, stuff like that.
[00:20:59.360] - Felix Krause
One random interesting find was the battery level. It's a very simple signal that usually doesn't mean too much. For one fitness app, we found that it decreased the conversion by a lot. I think it was 60, 70% or something. If the battery level is below, I think it was like 5% or something, but it was like a fitness app.
[00:21:19.330] - Felix Krause
You record your workout or something, you don't want your phone to die while you do that. Then we have another app that's more focused towards Gen Z, very young users, and turns out they really couldn't care less about battery. It's literally completely flat. That's why we have this calibration phase to calibrate our signals, and that's how we achieve the best performance.
[00:21:41.840] - Jeff Grang
Great. There's one thing I'd like to talk about, and I'm sure, I'm sure you've been facing the same issues. When you integrate an SDK, an external SDK, there's marketing team to say, "Hey, I want you to insert that new ad SDK within your app." You uncover the things and say, "Oh, wow, they're using this and these APIs and the SDK is really bad." It's always a fight between marketing and tech, and we always lose as tech people.
[00:22:10.550] - Jeff Grang
I'm sure every developer listening to the podcast right now have faced this situation. I know that when we build it, the first version of Purchasely, we really wanted to develop an SDK that we would love using and integrating, so easy to integrate, well-documented, lightweight, respecting the privacy of the users, et cetera. I'd like to hear your thoughts on this, Felix, because I'm sure that you've pushed that really far with ContextSDK being a good citizen.
[00:22:38.780] - Felix Krause
Yeah. As you mentioned, I think you and I have a very similar view on this and opinion on this. ContextSDK also is 100% written natively. It uses like, I think it's 700 kilobytes added to your binary size, which is really nothing compared to what you describe around some advertising SDKs using, I think, up to 100 megabytes or something.
[00:23:01.400] - Felix Krause
Very little CPU usage when accessing it. I think it's like 0.2% or something, like battery usage and CPU usage, bringing that down to basically nothing was a very satisfying and fun challenge, I would say.
[00:23:15.700] - Felix Krause
I think it's so important because if you provide an SDK, then it's going to be used on many tens of millions and more devices suddenly. It has huge impact. I think you always want to be a good citizen for multiple reasons. One is towards Apple, which is obviously the platform, but also towards users. You're just using up the battery or whatever you're doing. That's very much a big focus for us.
[00:23:39.460] - Olivier Destrebecq
If our listeners want to learn more about ContextSDK and how to use it and what it can do for them, where can they go?
[00:23:46.650] - Felix Krause
We have contextsdk.com. Right now we are not self-serviced, so it has to be going through email right now to onboard you because also we have for every app, there is going to be some different integration.
[00:24:00.990] - Felix Krause
For example, if you have paywalls and you have a paywall on the first app start, it's going to be a different integration than if you're just upselling something sometimes. We help you get that on board. Eventually, obviously it should be fully automated. Right now, automating something where it still needs some hand-holding and help from us and some guidance from us wouldn't make sense.
[00:24:19.580] - Felix Krause
Yeah, contextsdk.com, and then we have a contact form. We just fill out basically number of active users and the name of your app. The reason why we're asking for the number of active users is because our models are going to work differently. The more users you have, the faster you're going to have your own custom model and the more precise that model is going to be.
[00:24:37.410] - Olivier Destrebecq
Nice. If people want to learn more about you, I actually found out that you're very active on social network, but you also have your own website with a bunch of stuff about you. Where can people learn more about you, Felix?
[00:24:51.400] - Felix Krause
I have a few starting points. One is krausefx.com, where I have my blog and some information about me, as well as my privacy publications on the iOS platform. Then I have this other website that I haven't worked on recently, but it's a big website which is called How is Felix Start today? It's a huge quantified self-dashboard.
[00:25:12.940] - Felix Krause
It contains all this information about anything from the number of steps I took to my energy level, happiness level or how many hours I was in meetings. It's How is Felix Start Today? It started out as a fun project to just show where I am because initially it was called Where is Felix Start today? I was just showing my flight schedule for conferences and stuff that I spoke at.
[00:25:33.580] - Felix Krause
Then it just escalated up to be a full quantified self-dashboard, which was always like a hobby of mine. I'm still recording all of these things every day now, but I haven't updated the graphs yet. The graphs are always snapshots because I still want to make sure that data that's included there is not too personal.
[00:25:49.100] - Olivier Destrebecq
It's actually interesting that you mentioned the website because I went and read it and the conclusion there was it was a great experiment. I learned quite a bit about myself, but it was a lot of effort to do it. It's very interesting to hear that you are actually still doing it and putting the effort in, so it's great.
[00:26:04.120] - Felix Krause
Yeah. I think nowadays I would do a different day. Apple has launched some really cool Apple Health stuff and even the Journal app, which I haven't had a chance to try yet. I remember that Apple Health app now has the mood tracking building, for example, with a very nice use interface and everything. Maybe I'll build something on top of that instead if I were to build it now.
[00:26:23.320] - Felix Krause
The main conclusion was that you mentioned now was around the effort around building the data analysis tools because you need to set it up that you can even build those graphs.
[00:26:34.770] - Felix Krause
To be honest, nowadays I would have way better tools and experience thanks to ContextSDK because it's the topic of data science. Back then, if you don't know the tools, like if you as an iOS developer don't know how to use Xcode and code signing, you're going to have a hard time. That was me back then not having any data science experience back then.
[00:26:52.760] - Olivier Destrebecq
Nice. Well, thanks for joining us today. It was really nice to get all your answer and your perspective on your experiences and see where you're heading with ContextSDK. Thanks for joining us today, Felix.
[00:27:02.570] - Jeff Grang
Thank you, Felix. I hope we're going to have the opportunity to test both our technologies on the customer and see how both context and context trigger and contextual paywalls and experiences can leverage a better growth and our LTV in subscription area.
[00:27:19.300] - Felix Krause
Perfect. Thank you for having me.
[00:27:20.670] - Jeff Grang
Yeah, of course.
[00:27:21.920] - Olivier Destrebecq
Thanks, Chris.
[00:27:23.050] - Olivier Destrebecq
On behalf of the Purchasely team, thank you for listening to the Subscription League podcast. If you've enjoyed what you heard, leave us a five-star review on iTunes or other audio platform. To find out more about Purchasely and how we can improve your subscription business, visit purchasely.com.
[00:27:39.900] - Olivier Destrebecq
Please hit subscribe in your podcast player and don't miss any future episodes. You can also listen to previous episodes at subscriptionleague. com. See you soon.
