Crowdbotics Logo

Customers arrow

Don’t take our word for it, see what our customers have to say.

About Us arrow

We are on a mission to radically transform the software development lifecycle.

Home Blog ...

Case Study

Aura Health’s A/B-Driven Journey From Angel to Scale with Crowdbotics

Anand Kulkarni, CEO of Crowdbotics, sits down with Steve Lee, founder and CEO of Aura Health, to discuss A/B testing, consumer-oriented KPIs, making the leap into entrepreneurship, and using Crowdbotics to scale.

6 October 2021

by Crowdbotics

Anand Kulkarni, CEO of Crowdbotics, sits down with Steve Lee, founder and CEO of Aura Health, to discuss A/B testing, consumer-oriented KPIs, making the leap into entrepreneurship, and using Crowdbotics to scale.

Aura is a mobile app that anyone can use to reduce anxiety and improve sleep in just three minutes. Aura uses deep learning to recommend content such as guided meditations, stories, and life coaching personalized to you and your emotional state.

Aura was started two years ago. Today, it has several million users and is a frequent top ten in the App Store Health & Fitness category.


Anand: I’d love to know a little bit about the early journey for Aura. How did you come up with the concept and get the application off the ground?

Steve: My passion for mental health started when I was really young. My brother, who’s my current co-founder, and I grew up seeing our mother go through depression.

It was really shocking to us that, even as a doctor back in Korea, she didn’t really know what to do.

With that in mind, we imagined a solution, like a mobile app, that anyone can turn to just to simply share whatever they’re going through and really have confidence that this app will be able to help you.

My brother dropped out and I quit my job. We thought the timing was right. And we felt very confident that if we didn’t solve the problem then no one else would. So, I decided to jump right in.

In the beginning, it was myself coding most of the application. I’m not a traditional software engineer, but I did have some of the skill set. So, the intial version of the app was super lean. MVP style. We recreated the app within a few weeks and just released it. We didn’t make a big fuss about it, but started to get users slowly.

From there, we started running as many A/B tests as possible to prove hypotheses. We ended up creating a marketplace where therapists and life coaches and meditation coaches can start creating content and make money on our platform as well, just by helping more and more people. Until a year ago, it was just my brother and I grinding for a year. We got a very small check from angel investors and right away, we ended up working with Crowdbotics to scale up our engineering team.

As a startup, iteration speed is everything. We assume that we almost know nothing until we test it. We still run tens of A/B test per week and Crowdbotics has been a really strong research partner on that front. Right now we’re using two engineers  from Crowdbotics full time. And we plan on using much more in the future. We  raised a Seed round maybe four to six months ago and we now have six people working in person full time in San Francisco.

Steve Lee, CEO and Founder, Aura

Anand: Fantastic! As the founder of Crowdbotics, I love to hear when entrepreneurs are successful getting to scale on our platform. One of the things that is really impressive about Aura is the way you have maintained a very disciplined approach to making decisions based on data, as opposed to just intuition. How do you guys decide what you’re going to test? How do you actually run a test? And how do you know if a test is working?

Steve: At a high level, we have big hypotheses about where the product should go, where the market is headed, and how our product fits into that landscape. So, we kind of define these testing “themes” and we dive into those. We make guesses and design a bunch of A/B tests that prove or disprove our hypotheses. We look at everything as a funnel and examine where our weak links are. I recommend a blog called Reforged for more on testing strategy.

So for our MVP or MLP – “minimum lovable product” — whatever people call it these days, we really did not focus on making things perfect. We’re testing just enough so we can prove or disprove whether what we’re thinking is right or wrong. Sometimes it doesn’t go the right way. I would say  80% of our tests are failures. Sometimes we get interesting results, but we need to keep continuing to test and tie it back to the big idea or the theme so we can move the product forward. I would say that a lot of people, including ourselves, make the mistake of building too many features.

Anand: That’s really impressive. This kind of discipline is sometimes hard to come by. As you point out, features that do not work drive your overhead up and slow you down. That’s something  a lot of product teams, and especially first time founders, don’t recognize. How has your strategy changed as you’ve grown from just a few to few million users?

Steve: I would say our strategy is fundamentally the same. But as you grow, there are additional KPIs you need to start looking at. In the beginning, when you don’t have enough info to run all the A/B tests you’d like, you have to really rely on intuition. Accordingly, we considered qualitative feedback more highly. Right now we’re at a stage where we can rapidly run A/B tests and we rely much more on quantitative data compared to when we’re starting off.

The downside of having more users is that there are a lot more KPIs to consider. And sometimes balancing all those — let’s say, user engagement, growth, retention, activation — is going to be really challenging. It’s extremely important to focus on just a few things that really drive the business forward and make a better product.

Anand: As a primarily consumer-oriented app, what are some of your top key performance indicators (KPIs)?

Steve: Everyone would say you should focus on retention first. I agree with that. However, we did not actually put retention first. We really focus on activation. If you have a subscription or if you run paid ads, activations lower CAC and increase LTV at the same time. Even mature companies like Pinterest still heavily focus on activation.

Anand: Tell me about the technology stack. How are you guys constructed? And has that changed at all over time?

Steve: We’re mostly JavaScript-based on both frontend and backend. As we mature, our backend is getting much more complicated, of course. We have Python and different databases, systems, and jobs running. On the frontend, we stayed with a JavaScript framework, even as a mobile company, mainly because we could iterate with it much faster than using native platforms. We also went with Javascript and React Native because we wanted to support dual platforms with very limited engineering resources.

Anand: And then behind the scenes, Firebase, AWS, something else?

Steve: We’re on the Google platform at the moment.

Anand: Can you give me an example of something cool you guys have built with Crowdbotics?

Steve: If you download the Aura mobile app (iOS, Android), the majority of it is built using Crowdbotics. Of course, we have in-house engineers like myself who oversee everything and work closely with the Crowdbotics engineers, but a lot of the work right now is actually just built completely with Crowdbotics.

Anand: That’s terrific. It’s always awesome to see that we have such impactful uses cases like this with Crowdbotics. I think it’d be really cool to also hear a little bit about some of the inspirational success stories from people who are using Aura. Do you ever get people writing in to say, “Hey, this app has really changed my life for the better?”

Steve: Oh, yeah. We get it almost every day. We have automated emails that go out to our users from our co-founder, and people send back all sorts of great stories. Where people really see the most value is just regaining a moment for themselves when they’re really busy with this modern world and kind of losing a sense of themselves. This manifests as stress, anxiety, insomnia, all those. Our product basically uses machine learning so that you can have the best quality time for yourself. It’s been incredible to hear from people who couldn’t sleep – who just couldn’t bear their lives – and are finally building healthy habits over time with Aura. Those are the most gratifying stories we hear.

Anand: Any final advice you want to give to folks who are thinking about building a new application, or advice you might have given to yourself two years ago when you were at the beginning of this process?

Steve: I think if you truly believe in the problem, and if you also truly believe that you’re the right person to solve the problem, then you should go for it. I would say, just don’t be too afraid. Take your time as much as you need but if you feel ready, then go for it and keep your head up. It’s going to be hard. Entrepreneurship is much harder than I expected before jumping into it. But that’s made it all the more gratifying.

To learn more about Aura, visit aurahealth.io. Aura is available today in the Google Play and Apple App Store.