What's Your Life About?
Paul Kemp: Welcome to another episode of The App Guy Podcast. I am your host, it's Paul Kemp. This is the podcast where we really do appeal to app entrepreneurs, app founders, startup founders or if you're working on a side project. In fact, anyone who wants to really make a difference in the world of tech. I have a lot of archive episode. To listen and access to over 500 episodes of The App Guy Podcast, either search Paul Kemp in any of your podcasting apps or click Archives on my website http://TheAppGuy.co.
This is episode 513. Today, I'm really excited about my guests I have on the show. They're based in San Francisco - the hub of where all the tech action is happening. They are into app intelligence. Let me introduce Andrew Levy and Robert Kwok, the co-founders of Apteligent.
Apteligent is an app intelligence company (based in San Francisco), and it is software which allows you to get app insights into events such as: app crashes, app load times, network latency etc. These co-founders have also fascinating startup stories, so let's get stuck in.
Andrew, Robert, welcome to The App Guy Podcast.
Andrew Levy: Thanks for having us.
Robert Kwok: Thanks for having us.
Paul Kemp: Thanks for coming, both of you. Let's jump straight into you, Andrew. I would love to know how did you get into the startup scene?
Andrew Levy: Sure. I was going to school at a college out in Baltimore, and I knew I wanted to start a company early on. I did all the usual entrepreneur classes and business plan competitions in college. I actually got first place, so that motivated me a bit. Then I knew I needed to be out in the Valley - or wanted to be out in the Valley - to eventually start a company.
I got a job with HP that offered to pay for me to move out here to the Bay Area. That sounded great at the time. I flew out, I started working there... That's actually where I met Rob. We were working on data warehousing systems there. To be honest, it wasn't the most exciting job in the world, but we learned a little bit about enterprise software, and I guess eventually the kind of company we wanted to create. We wanted to create a different type of company than the one we were at at the time.
Paul Kemp: Were you both working in a garage, by any chance, like HP started?
Andrew Levy: No, that would have been fun... It was more like a giant cubicle farm. [laughter] But we met a lot of great people. We actually went our separate way at this point. Rob started a different company - he can talk about that - and I got into Ycombinator, a startup incubator based out here. That was about 18 months into HP. After I got in, I quit. The other two co-founders at the time were at Stanford; that was for a different company than Apteligent. We were focused on applying collaborative filtering to e-mail. Essentially, you can think of Gmail Inbox before that existed.
We found it was very tough to monetize e-mail. Eventually, we got to be working on a few other projects, but didn't quite become passionate anything else we were working on, so by the end of 2009 - it was that summer that we were in the incubator program... By the end of that year, we just decided to disband that company.
We also happened to go through a different accelerator - Lightspeed Ventures - we did two at the same time, but it ended up working out, because after we disbanded that YC startup I met up with Rob, we started working on a few ideas together, eventually got to be building mobile apps and that's what lead to the founding of Crittercism, which was Apteligent's previous name.
Paul Kemp: Okay. Robert, in terms of your story - give us an insight into how you actually got into startups.
Robert Kwok: Sure. It's kind of a similar story to Andrew. I've always wanted to start a company. Just talking about it, I realize that it's been in the family business for a while. My grandpa moved here from China, and he actually started a restaurant near San Francisco, in the East Bay. That restaurant became successful, and that's how he brought the rest of his family to the U.S.
There's a crazy newspaper story about him and how he was separated from his wife for 20 years while he was starting this restaurant. My parents also started restaurants; my dad started one as well, so I guess it's kind of been in our family history there.
When I went to high school, I used to buy and sell magic cards, that was my side business back then. Then I went to MIT, and I guess I just met a lot of like-minded people there that also wanted to dream big and change the world, so I would spend a lot of late nights talking to friends about ideas in terms of what we could build and start companies - that was always the plan.
I started working at HP after MIT, and met Andrew there, as well. My plan was basically "Let's work there for three years, get some experience and then I'll quit and start a company." I think literally it was almost exactly three years after I started where I left HP, and then moved to Berkeley and started a company with some other MIT friends. Coincidentally, it was also an e-mail startup idea. I didn't know Andrew was working on e-mail at the time.
Our idea was a little bit different... We were trying to create a new interface to triage e-mails. You can imagine, e-mail's been the same for the last 20 years - a list of e-mails on a web page - so we had this idea where it'd be a 2D interface, you could have all your work e-mails in one corner, you could have different sizes and shapes; more people on the thread would mean the shape would be bigger, more important e-mails would be red, versus blue... That was kind of the idea, and we ran into similar problems where people just wouldn't pay for e-mail.
We also had a huge falling out in the company because two of us had a crush on the third co-founder, so that was a major problem. [laughter]
Paul Kemp: Which one of you won?
Robert Kwok: We actually both lost. She started dating this 40-year-old guy that she met on OkCupid.
Paul Kemp: Well, her loss anyway.
Robert Kwok: Yes.
Paul Kemp: Both of you - that's fascinating. What I've learned already, and from all these episodes I've done of the App Guy Podcast (over 500), one of the biggest takeaways is that entrepreneurs -- it's in the family. If you are listening right now or reading this and you have entrepreneurial parents or grandparents, just expect that it's in your blood.
Let's transition then into Apteligent. I'd love to know what you can do for the app developers, the startup tech people that are listening to this. How can you help them with app intelligence?
Andrew Levy: Sure. In the early days of Apteligent we very much talked with a bunch of other mobile entrepreneurs, mobile developers, mobile teams, and they were experiencing the same issues that we did while we were building mobile apps, which was that it was very difficult to understand user experience, especially once you released the app out into the app store.
You see those app reviews saying, "This app sucks, it doesn't work", it's really not helping you get to the bottom of what actually happened. So what we created with Apteligent is a data-driven approach to understand where users are struggling in the app, and most importantly why.
Commonly, a lot of people will talk about analytics tools. Most of those tools just look at user behavior and they'll tell you that one of your funnels - say checkout - is dropping on step 3 of 4. That's great, but they don't tell you why are you losing those users, what is happening to them. So what we did was we built an SDK, a tool that you can embed into your apps and it sends off real-time diagnostic data about the performance. Think about all the ways an app can fail: slow load times, crashes, network problems, it could be a device or an OS-specific issue, a location-specific problem...
We combine the performance data with the behavioral data, so what the user is doing and how performance impacted their behavior. Then we connect that to the business impact. For example, if you have a high crash rate, obviously your churn will be impacted. That's what's unique about what we do - our ability to connect a technical tool with the business impact.
Paul Kemp: Robert, I'd love to know the idea for this then - did you fall upon it based on the work that you were doing, or did you collaborate in a more strategic manner to bring this to life?
Robert Kwok: I guess there are a couple things. One, we were developing apps for ourselves, and we had an issue... There was actually a bug in some of the code that I wrote, where we'd worked three months building this app, and when we tested it everything was great, but then when we launched to a new user, it was looking for some value that would only get set if you're an existing user. So every new person that would download the app would have this crash.
Apple's App Store at the time didn't really have much there - I don't think it still has that much now - and it told us everything was fine, but then all these people started e-mailing us and saying, "Hey, your app doesn't work. Every time I load it, it crashes, so I can't even use your app." So we thought, "Well, there has to be a better way."
Paul Kemp: What's really interesting is that a few episodes ago we talked about the idea of building things for yourself and discovering problems. Scratching your own itch. How did identifying this problem lead to becoming a fully-fledged successful startup called Apteligent.
Andrew Levy: Right. So at this point, we had been iterating on a few different things for a while, so we were running out of money. We really wanted to make sure that this itch that we were scratching was real, so we did a few things: we spoke at a bunch of mobile meetups to hear feedback.
The other thing that we did was we created a simple landing page, a form where users could sign up, they could leave an e-mail address if they were interested in finding out more information, and we made sure that we hit some thresholds in sign-ups and combined that with the feedback we were getting to validate that we were on to something.
We also ended up applying to a different incubator called AngelPad. We had those sign-ups and that validation, the feedback, and then we actually ended up getting into that incubator, so a combination of both of those things - we said, "Let's go for it."
Robert Kwok: We also happened to know a lot of app developers at the time. For example, one of our friends actually made the number one gaming app in Asia at that point; it was a simulation cooking game. It had millions of users, and we were able to do a lot of the grunt work and just try to get as many people that we knew using it, going to meetups and talking to people there, going to hackathons, and just having that level of one-on-one connection with people really helped validate the concept also.
Andrew Levy: Right. The idea was to do unscalable things at the time, and I recommend everyone do those things.
Paul Kemp: What I'm learning, guys... You're picking up on many of the themes that The App Guy Podcast has touched over the years, and one is the importance of your own network and tapping into that, which clearly benefitted you with the game. Also, really trying to understand the feedback from the potential customers and users of your service. Do you feel like there are any other tips you can give any of those startup founders who are listening to this and want to give themselves the best chance of success?
Andrew Levy: Well, as I was alluding to before, I think people tend to hold their ideas to their chest, but like you said, you need that feedback, you need to get out there and talk about it, bounce ideas off folks, and part of the benefit of being in a place like the Bay Area is, like you mentioned, the network. It's not the only place to start a company; certainly, there's other startup cities all across the globe. It definitely makes it easier to be in those places where you can hang out with folks and talk about AI, or self-driving cars, or whatever you're into. Surrounding yourself with like-minded folks will lead to a higher chance of doing great things.
Robert Kwok: The other mistake we actually made early on was we would spend too much time perfecting something before releasing it, and when we switched to this new idea I think we just went for it. A lot of it is just getting over this fear that it's not going to be perfect, or it's not going to be what you really want it to be, and just launching it and see what happens, and just keep iterating over time.
Paul Kemp: Yes. So you are an app intelligence company; I wondered, do you actually aggregate any of the data that you get on the backend? For instance, the number of crashes, or the average load times. Do you create some reports or anything that you can share with us about the aggregate data that you're seeing?
Andrew Levy: Yes, certainly. I encourage you to check out data.Apteligent.com. We have a lot of publicly available benchmarks there, and also some reserved for enterprise customers. What we learned over time was that we're collecting a ton of rich information about the mobile ecosystem; let's leverage that in a way that we can give it back to our customers in the community in the form of industry trends and benchmarks, like average crash rates, average load times...
We have a global device directory, so if you're trying to launch an app and you're trying to figure out "What devices, operating systems, configurations should I be testing on? What is the average stability or bandwidth available in certain regions?" We provide all of that data so that you can create your test and release plans and you can benchmark how well you're doing relative to your competitors in certain app store categories. Tons of rich information.
We also - this is really interesting - have been able to apply some data science and machine learning to the data to understand how performance is impacting business KPIs. I'd mentioned churn earlier, and we actually found that crashes can decrease next-day app opens by up to 8x, and they can increase churn overall by up to 6x. So huge impact on just that one performance metric along.
Of course, if your other metrics are not doing too well - it's taking too long to load, or certain flows are delayed - it will have an impact on the business. On that site, you can actually find -- we publish research reports every 4-6 weeks that go into even more detail about the industry.
Paul Kemp: Yes, and just to ensure, for those who are listening whilst driving - you can go to theappguy.co it's episode #513, with Andrew and Robert, and you'll see links to that site. I'm also wondering... You've seen a lot of app developers - appreneurs, as I like to say - start business, who have their apps, and I wondered what you feel like the most important metrics that they should be looking at are, in terms of what you've learned over the years.
Robert Kwok: When you're developing an app, it's great to monitor how your app is doing over time. One of the most critical moments is really around when you first release an app - tracking the first 72 hours after your release. The metrics there that you really want to track are: how many people are adopting that version, how many people are upgrading, and then being able to understand why they are or aren't adopting it.
One of the metrics a lot of our customers track is the crash rate of their app, which is basically the number of sessions that end in a crash. That's often a very good indicator; it's highly correlated with app store ratings, which help drive downloads, so it's kind of a feedback cycle there.
The other metric a lot of companies are tracking today are app start time. Just thinking about it, if you're using an app and it takes ten seconds to load, you're probably going to give up at some point and stop using that app and maybe use a competitor's app. Those are two metrics that a lot of companies closely track, especially when they're releasing new versions.
Andrew Levy: And just to pile on top of that, those performance metrics impact your business, and I think what many first-time app publishers make mistake of is they focus on conversion rates in certain flows in the app. Those are obviously important, but they'd really need be focused on engagement and retention first and foremost, before they ramp up marketing campaigns... As well as those performance metrics that Rob mentioned, because what's going to end up happening is you're going to spend a ton of money to get users, they're going to have a poor user experience and you're going to have a leaky bucket. You're going to lose a ton of them, your acquisition cost will go up, and it just won't be a happy ending. You definitely need to make sure you optimize those metrics.
Paul Kemp: This is great, because it's really compounding a lot of the information we've got from the past amazing founders, and certainly I've had a first-time experience of what you mentioned as being the leaky bucket, where you've got tens of thousands of users downloading the app for the first time, but maybe the conversion is not right... In terms of a conversion rate, what do you think is a good conversion rate? From landing on the app store page to then clicking the Download button.
Andrew Levy: It's tough to say... It can be very vertical-specific. Frankly, our metrics don't focus on the attribution piece. There's tracking of once they open the app, but then there's also "How do you connect? Where are the users coming from?" and those conversion rates. We focus mostly on, you know, once the app is opened what are the metrics that you should look at.
There's some data out there - if you're interested in this topic, I would look up mobile app attribution of Google.
Paul Kemp: Yes. And as we draw to a close, I'm wondering what exciting stuff you're working on now, in terms of future updates or anything that's coming out going forward.
Andrew Levy: Yes, we continue to add new types of data science capabilities into our platform. We're really driving towards this idea of being able to say, "Hey, if you fix these three issues, then your churn will decrease by this amount, or your engagement will increase, or your revenue will increase." So being able to proactively recommend areas that teams should focus in on is a key area for us.
We also continue to add other performance metrics. We just launched UI hangs - or you can call it app hangs and out-of-memory analysis - into our product. We're also adding some custom reporting for enterprise clients... So a lot of new dashboards that teams can play around with.
Robert Kwok: We're helping companies understand the wider user experience of the app. So not just knowing when the app crashes, but when the user gets frustrated by the app, or when it's draining battery or killing your data plan.
Paul Kemp: It's really exciting, because I can imagine that you're also looking at artificial intelligence or machine learning or something, to understand where in the code the crashes are happening, and perhaps give developers more insights into the parts of the code that are crashing. Are you looking into that stuff?
Andrew Levy: Yes, we've done some of that already. We launched a product called Smart Crash Search, where we identify similar root causes across errors. One of the common problems that people have is there's a lot of noise generated from these apps, so you need to understand where should your team be investing their time and what are the related problems. So we actually give a probability score in this new product to help you understand how similar is this issue, compared to all these other issues. It's a really interesting and unique capability of ours.
Robert Kwok: We also launched something recently called the 3rd Party SDK Analysis. One of the common problems when you're developing a mobile app is to figure out which SDKs you should integrate. Often times those will cause performance issues like crashes or slowdowns. We launched a product a couple months ago that allows you to recognize if a crash is actually caused by a third-party SDK.
Paul Kemp: Fascinating. Sadly, we've got to draw to a close now... But I'm wondering, do you feel like we've touched on everything in regards to what you wanted to talk about Apteligent? Is there anything else you want to add for those who are thinking about using you?
Andrew Levy: One thing I'd say is we have a free tier as well; we do work with some of the largest app publishers across the globe, but we also work with folks in a garage somewhere, looking to build the next big app. So we have a free tier, we also have a middle tier called JumpStart for small businesses, and our enterprise tier. I'd encourage anyone that's interested to check us out. It's very easy and quick to get started, at Apteligent.com
Paul Kemp: Wonderful. Andrew, Robert, thank you both for coming on The App Guy Podcast and sharing your story about Apteligent and what you can achieve with the software. I highly recommend the appster tribe go and check you out. How best can people reach out and connect with you both? What is the best way of getting in touch?
Andrew Levy: If you e-mail info [at] Apteligent.com, it will be sent out to us. Or you can just contact us directly - Andrew [at] Apteligent.com and Rob [at] Apteligent.com.
Paul Kemp: Great.