Hello! What's your background, and what are you working on?
My name is Alan Warsoff, and my background is in marketing. I started my career working at a digital marketing agency, followed by the marketing department of a video game company, where I made the transition to software development.
I crashed a company hack night and built a machine learning algorithm to automatically optimize ads. I was asked to present it to the department, and they moved me to a more development oriented job. Later, I left to make SelfDecode.
SelfDecode is a genetic interpretation tool and a personal digital health coach. We have algorithms look at your genes and give optimal recommendations based on your predispositions.
We launched in June 2016, and today we have 9000 users and are making $17,000 per month. We have three people working from a house we rent in Los Angeles, and we've also opened an office in Ukraine with three developers.
How did you get started working on SelfDecode?
My cofounder, Joe Cohen, runs a popular biohacking website called SelfHacked. Through consultations with his clients he discovered that people wanted genetic interpretation, and all of the available products were inadequate, so he began working with a developer part-time to create the very first prototype of SelfDecode.
Joe and I became friends online by exchanging biohacking studies involving the mechanisms of memory and intelligence. While still at my job at the video game company, I started working remotely to help with whatever I could on SelfDecode while it was in beta.
At the time, I knew that I wanted to start a business and was toying with a lot of different startup ideas. For example, an online school that revolves around spaced repetition, some marketing automation tools, etc. However, when I saw how engaged the beta users of SelfDecode were, I decided I should pursue that idea full-time.
I moved in with Joe in March 2016. This was a big risk, because we had never met in person before, and had only had a handful of conversations. But I figured that he seemed like a reasonable person, and I tend to get along with reasonable people. Neither of us had much money, so I moved into the walk-in closet of his bedroom while he rented out all of the other rooms of the house to cover the bills. I lived in that closet for 6 months.
What did it take to get the product off the ground?
After I arrived, I urged Joe to launch the product as soon as possible so we could become a viable business. But, justifiably, Joe did not want to release the program in its broken state. The first prototype was so buggy that users couldn't even sign up half the time. People would complain about getting multiple billing receipts thinking that we charged them multiple times.
We made the decision that the current dev team and framework we were using (WordPress) were not conducive to an agile and robust product, and we could not afford to hire experienced developers, so we decided that I would rebuild the whole app in Django. I chose Django over Ruby on Rails because of its data science libraries (like Pandas and NumPy), which came in handy processing user genome files which had a half a million rows each.
Joe had begun working with the first developer part-time in March 2015, and they'd spent about a year on-and-off working on the app using WordPress. When we told the original developer that we were rebuilding the whole app, he was sure it would take more than 6 months to do.
Luckily, my core skill in development is speed — I built the whole app in a bit over a month and we were finally able to launch bug-free soon after that. During this time, Joe and I continued to work on SelfHacked.com trying to increase revenue to fund SelfDecode.
What's your tech stack?
We use Bootstrap, jQuery, Python 3, Django, Postgres, Heroku, and some key Django apps including django_extensions, django_hijack, django_allauth.
How did you find your first customers? And how did you grow from there?
We launched in June to the SelfHacked email list, and we immediately made about $300 that day, which to us seemed like a ton. Today, 60% of SelfDecode's users still come from SelfHacked.
The second biggest source of traffic diversion is via embedded links in the post content itself. Whenever a post lends itself to genomics, we talk about the genetic mutations that are relevant and link to SelfDecode for more info, or to allow the user to see which mutation they actually have.
One thing we learned was that the small details matter when it comes to the performance of an ad. In one instance, we took an ad stretched horizontally across the nav bar, changed the button color from blue to yellow, and configured it to remain fixed to the top of the browser as people scroll. This increase its output by almost 4x.
Where does the other 40% of your traffic come from?
The rest of our traffic comes from Google searches. Similar to SelfHacked, we've focused on creating content that will get ranked by search engines in order to get traffic. My advice here is to focus on content marketing, since paid acquisition is too unprofitable for products that have not yet been optimized.
We are biohackers ourselves, and so most of the time we simply write posts about topics that we are interested in. We also ask ourselves the question, "Does the internet need a good post about x?" It helps to look for topics that our competitors are writing about, and then write about a similar topic when we see their post is lacking valuable information.
An important part of SEO and content marketing is to optimize conversion rates, so that new visitors actually become customers. One way we increased registration conversion was by ruthlessly eliminating unnecessary fields from the registration form. Right now we have two fields, and we hope to bring that down to one.
How does your business model work? What's the story behind your revenue?
SelfDecode currently makes $17,000 per month. Our main source of revenue is through our $19 per month subscription.
We started charging on day #1. Given all the bugs, we only charged our beta users for one month. We started out using only a Stripe Checkout form, but we got a lot of complaints from users saying they only trust PayPal, so we added PayPal a few months later. As a result, checkout conversion went up a lot (~20%).
Our expenses have skyrocketed since we began employing a dev team in Ukraine, and so we currently make no profit. The rest of our expenses (Heroku and AWS) are trivial, and Joe and I are introverts and spend very little money for living expenses.
My advice would be to do some research and understand all of the different monetization strategies available before you choose one. These include sale of digital products, affiliate links, ecommerce, drop shipping, subscriptions, advertising, and others. Don't just pick one blindly without looking into it.
What are your goals for the future?
We want to have the most comprehensive recommendation engine in the world. Upcoming key features include the integration of blood tests and symptoms into the system, so people will be able to upload more and more physiological data and get even more accurate and efficient recommendations.
We also want to grow the business and keep hiring, so we're adding developers and designers as fast as we can afford them.
What would you do differently if you had to start over?
The biggest mistake we made was choosing WordPress as an app framework and relying on a developer who was creative but did not execute tasks well. People have different abilities, and we needed someone who could take an idea to completion. If I were to start over, I would have worked harder to make sure the person we hired had execution skills.
Another mistake we made was not researching the people we met in Silicon Valley. We traveled there twice, and both times we met people who seemed legitimate but after some investigation turned out to be posers and frauds, which was a huge waste of time. We did also meet one great advisor there, but the vast majority of meetings were a waste of time.
What were your biggest advantages? Was anything particularly helpful?
Our biggest advantage is in hiring. Throughout the course of hiring for SelfHacked and SelfDecode, Joe developed a system for filtering through job condidates by administering a barrage of cognitive tests for various cognitive abilities. Through backtesting on our current workforce, he learned to associate performance on certain tests with certain types of job performance later on.
This has saved us an enormous amount of time and money on training and rehiring. Then to free up his time he trained a virtual assistant to administer the tests instead of him. For us, the key to a great company is hiring great people, and that's very hard to do reliably without a stringent protocol.
My main advantage is that I am a generalist I know enough design, marketing, and software development to never need anyone else's help to build and launch a product.
Another advantage is that I'm good at teaching myself skills, so any time I feel I'm lacking in any skill that is holding me back, I just learn it. I use spaced repetition tools I built for myself to accomplish this as well as nootropics. My general strategy for learning a new skill:
- Rote memorize the definitions of all of the jargon. This helps when things are being explained so I don't get lost in the jargon.
- Write down every piece of knowledge in my own words. This forces my mind to clarify the ideas.
- Do it every day. This forces my brain to get into learning mode.
Another big advantage, of course, was access to the SelfHacked audience, which still accounts for half of our revenue.
What's your advice for indie hackers who are just starting out?
I recommend reading Zero to One by Peter Thiel, The Lean Startup by Eric Ries, and The 4-Hour Workweek by Tim Ferriss. I think all three of these books teach important lessons in very different aspects of startup building. They sometimes contradict each other, because they are different philosophies, but I will list my key take aways from each:
Peter Thiel is a talented contrarian and thinks from first principles. While it is clear that he exaggerates certain ideas for dramatic effect, he offers some key insights that are counter to the Silicon Valley kool-aid. As startups become a more fashionable thing to do, there will be an increase of people who are not critical thinkers trying to start companies.
Keeping with Peter Thiel's style of thinking (thinking from first principles instead of analogies) will protect you from the pitfalls of groupthink. If your business idea is intuitive, there will probably be a lot of competition, and to come to unintuitive truths requires thinking from first principles.
The Lean Startup is a bit wordy, but establishes a framework software companies can use to quickly adapt to customer needs.
The 4-Hour Workweek contains many ideas that you can take or leave, but Tim puts forward ideas about productivity that I have found particularly useful. For example, asking yourself introspective questions, batching, and placing yourself in an environment conducive to productivity rather than relying on the power of will.
One mistake I see people make is underestimating the innovation required for sales. A maxim I made up is, "An innovative product requires equally innovative sales."
Another mistake I see people make is not understanding the business as a whole. Many people will focus only on technology and will face difficulties when it's time to execute the other parts of the business.
I also follow the maxim, "capability comes before opportunity," which means that depending on the skills in your toolkit, you will be able to see more or fewer opportunities. Had I not taught myself to code, I would not have been able to build the app. Had I not taught myself marketing, I would not have been able to monetize it. It's meant to encourage me to pick up skills that are open-ended and useful.
Some key skills I would encourage others to learn the 80/20 of are VR, 3D printing, software engineering, electrical engineering, UI/UX, graphic design, hiring, application frameworks like Ruby on Rails and Cordova, data science, machine learning, analytics, marketing, and Excel.