How I in-validated a business idea in 3 months with little code, while traveling the world ✈️
In this article I’ll share with you how I validated a fresh idea, by focusing on metrics instead of lines of code; did that in 3 months, instead of 2.5+ years (like in my previous startup); got accepted in Y Combinator’s Startup School and… all that while traveling across San Francisco, Singapore and Bali. Buckle up!
It was mid-April this year, I was in SF, recovering from my previous startup failure, trying to meditate and staying away from jumping on new ideas, when all of a sudden… well I again got hit by a new idea! “Danger Danger!“ — my mind told me — “avoid past mistakes you should. Give it time and space, you should, before pouring efforts in it”. Shuuuure…
I was going through a large list of newsletters that linked to an even larger list of articles. Reading some, Pocketing others… and you know, never getting back to Pocket to actually read them. 😅 Same time I started listening to podcasts again. And on one of the monday mornings it struck me — what if I could save an article, the same way I could with Pocket’s browser extension, but it gets converted into an engaging audiobook! Not a robotic-sounding text-to-speech, but something that’s actually read by a human. That was it!
Again, I wanted to avoid mistakes I’ve done in the past and do hell-lot of research on this one, before sinking in too much energy and being attached an idea that won’t work.
Is there demand?
After asking a few friends what they thought about the idea, I quickly saw two groups emerge — those who were ecstatic about podcasts, audiobooks and spoken comtent… and those who didn’t care at all. Surprisingly I found zero people in between. That was actually pretty good, I thought! That’s the reaction a good idea or product should get. Hate it or love it! 😤 or 😻! Great ideas are polarising. I was surprised that audio content was like that. I felt I was onto something.
Next I took a look at some macros and comps. What’s happening to podcasting consumption and whether anyone has build anything like that.
- Market is growing and got ton of potential… In the USA, that is…
- YC had a podcasting app in their W17 batch called Breaker — another good sign!
Can this be done?
Okay. So I learned that there might be more than 1 person (me) who wants this and likely to use it. Next on, I wanted to figure out how to deliver the service. I saw 3 routes:
- Speech synthesis (TTS)
- Crowdsource narration from students
- Hire professional voice actors
I started by evaluating the state of speech synthesis tech. I tried out all available TTS solutions for the start. The one from AWS, that’s powering Alexa, one from Google and a few others. As I expected all of them sounded too robotic and far away from producing engaging long-form spoken content. Like audiobooks or podcasts.
Price: cheap 🤑. Quality: not engaging 😒
Next, I tried narrating articles myself. Ableton Live + standard Apple earphones with built in mic. I’ve been uploading my audios to SoundCloud here and sending samples to friends.
Quality was variable. Bearable, I’d say. Far away from sounding like a pro.
I did this for a week. Every day. One surprising side-effect that I found, is that for someone like me, with english being second language, it’s an amazing way to practice pronunciation and boost confidence with practice (perhaps another startup idea).
Price: cheap to mid-range 🤷. Quality: may vary 🤷
Okay. Next, I decided to find out how much a semi-professional and professional voice actors could cost. Long story short, it was anywhere between $10/hour (if you are lucky to find a good voice for that), up to $100/hour. Averaging 40–50 bucks for a quality North American accent. Source: personal network, Fiverr, Upwork.
Price: expensive 💰😱. Quality: high quality engaging content 😎
But the voice quality is clear. Compare my voice, you’ve heard above, with a professional. Here is an example, narrated by Darren Marlar. Darren rocks 🤘!
Considering such high costs for content production, we better start off with a category or a submarket with a high content reuse.
Research Conclusion 📚: I decided to rely on professional voice actors. First version of service should “wow!” early adopters.
Current progress: nothing built, no name, no domain name, no landing page, 2 weeks of casual research. Excitement builds up 😁 but my mood is mostly chill so far.
So far all this research might sound boring. Loads of talk and little action — right? Especially for someone with an engineering background like myself, who jumps on coding a full-fledged backend and native mobile app right on… “because only after the product is built you can validate its market potential” — no no no! Bullsh*t! In most situations you don’t have to build anything, to validate demand. This is planning! Remember what Winston Churchill said?
“If you fail to plan, you are planning to fail.”
YC Startup School
I got accepted into YC’s Startup School! OMG! Funny enough I applied with my previous company, that I just “closed” after 2 years of hustle (follow me, I’ll blog about it some time soon).
Second week of Startup School I told my mentor, Brad Flora, that I’m changing the direction. His feedback was: “yes! it’s a great idea and a growing market! In fact, just last week I invested in [company in a similar space]…” I was like: WOOOW! The reaction was 10x compared to all reactions that I ever got to my other projects. Everyone in the batch loved the idea too. I was ecstatic! 😃
Next day, my emotions were 180 degrees. Learning from mistakes of my previous project I wanted to monetize this ASAP. Current idea was consumer focused, and it would take years to make a dime. I put on my smartass hat and decided I should do a B2B detour to break-even first, and then, at scale, go back to target consumers. Holly f*ck this was a bad idea. It was driven by fear and previous negative experience. So I went on for 3–4 weeks to research potential B2B avenues for “article to human read audio” service. That journey was not bad — it was a roller coaster 🎢.
First I thought about narrating content for teams. Soon, it turned out that reading white papers is a “thing”, and teams got to do a ton of that. I shared this idea on one of the weekly group calls and then this happened:
I felt ecstatic again! 😁
Shortly, Kris Foster, fellow Startup School batch participant, introduced me to an HR/culture manager from Cisco HQ. YC people rock🤘! Had a call. They could use our service to let their engineers listen to lengthy whitepapers on the way to work. Sounded great! Quickly, I realized that technical whitepapers are also hard to narrate – due to figures, formulas graphs, and charts. You just have to see them to understand. How much they were willing to pay also didn’t sound exciting and barely covered production costs. But I was still quite high and excited about the progress!
Some tangible progress. Finally!
One more week. Only now, I finally decided to come up with a name, domain and a landing page.
ReadByHumans.com — boom 💥 Thank god domain was available!
Simple landing page built on Sails.js, Mongo to store email sign ups, deployed on dokku on AWS Lightsail. Nothing fancy. Took 1 day. Just used the tech that I use on daily basis. It’s tempting to try out newest shiny tech for your new business idea, but you are just adding risk, instead of mitigating it. It was tempting to go Serverless + react-navite-web + redux + GraphQL+ god-knows-what.js. But no! Resist the temptation.
rm -rf your engineering pride. Validate your business, not your tech!
🇸🇬 Singapore. Roller coaster is going down🎢 😧
End of April. Goodbye SF. Had to fly back to Singapore (That’s where I live). Startup School is still on. I went out and shared ReadByHumans idea with friends, fellow entrepreneurs and VCs. I was surprised how different the feedback and environment was compared to the Valley! Suddenly majority is skeptical and no one really listens to podcasts in Asia. In the USA, meanwhile, podcast consumption has been growing something around 20% year on year. South East Asia — flat. Ouch!
Why podcasting is a big thing in the USA? My hypothesis is, it’s because of the high car 🚙 ownership in the USA, compared to SE Asia. In the USA everyone is listening to radio / Spotify / audiobooks while commuting to work.
While in Singapore, I tried out another spin on B2B. Targeted companies that rely on content marketing. My sales pitch sounded something like “convert your written content & case studies into audio, so leads could consume it faster and be sold, faster.” After talking to a few SaaS businesses, I quickly learned, from their lukewarm reaction, that the idea was meh.
Product would also require additional operational and feature overhead — analytics, conversion tracking, consistency of voice actor’s voice, ensuring voice actors are not reused by competitors, and so forth… Alrighty. Moving on… 😒
I was frustrated with myself more than with challenges of selling B2B. Suddenly I felt that these 3 weeks I was not solving my own pain. I was trying to please some companies, teams… I almost forgot how and why I started this project. If you are to pour your sweat and soul into something, you better be solving your own pain and find other people who share this exact same pain with you. The time was ripe for yet another change.
🌴 Bali. Getting real sh*t done
After spending two weeks in Singapore, I went to Bali for a month. I wanted to focus on ReadByHumans and ship something by the end of Startup School. There, I caught up with my friend Andrey Azimov, who recently exploded with his When To Surf. We thought it’d be cool to work side by side and keep each other in check. He also suggested to get my sh*t together and to ship, measure, rinse, repeat. 👊
Whole of June was dedicated to make and test an UVP (unique value proposition). I decided to build a simple landing page with an email sign up, then measure what conversion rates I would get on all steps of my funnel, then… But first — UVP.
Clear title and call to action are critical. And basic imagery to convey what ReadByHumans was all about. For this launch, I had to go for a compromise. I wish I could offer “We narrate any article on demand for free”. But that would just bankrupt me. Hence, I decided to narrate top 3 articles from Hacker News. Top 3 articles for FREE upon sign up + 1 more article weekly. That value prop was concise, it was a no-brainer “yes”, and I could afford to pay my voice actors for 1–3 articles per week. For a few weeks that is. Not forever.
A little bit of magic and the final result is:
The landing page is done. Email capture is ready. Autoresponder with top 3 articles is set up.
Time to Ship! ⛵
I had an external deadline, which was YC Startup School Demo Presentation. June 16th. I aligned my Product Hunt and Hacker News submissions to be around the same dates.
I followed best practices of Product Hunt launch. Tons been written about that, so just Google “Product Hunt launch checklist”. Tuesday midnight, SFO time. Which is 3 PM in Bali and Singapore — no need to stay up all night! One way or another, you still should be ready for a 24+ hour sprint with minimum sleep. Handling server load, replying to comments, calls, etc. Do take time to prepare. Following up on all the comments is something special! It’s a race!
Chris Messina and Ryan Hoover jumped into the conversation. My heart was beating!
You can read the nitty gritty of the convo in the Product Hunt post here.
YC Startup School Demo Day was on June 16th. We got some traffic from there too. Also got good feedback. Notably, YC Partner Adora Cheung (Co-founder & CEO of Homejoy) tweeted about ReadByHumans all of a sudden:
And here is the application video itself:
Startup School launch felt like a complete success!
I had no idea how to launch on HN. Andrey had no idea as well, so he ran this by Pieter Levels. Pieter said “it’s a mystery, but it’s key to balance upvotes with genuine comments”. Oh well, let’s give it a try. As you will see below, most of the traffic came from HN, but not the most sign ups.
On HN, my “Show HN” post quickly became a controversial topic. I did not anticipate that, to be frank.
I think the very first comment was raising copyright implications.
It’s a long and complicated topic and I’m also not a lawyer. You can read the debate youself, if you’ve got time.
One thing I learned here, is that it’s illegal (in most cases) to make a derivative work sell it via distribution. Distribution, basically, means re-packaging and reselling someone’s content. I found two ways of solving it:
- Strike deals with publications and authors. But I could not do that in an individual capacity, since it’s just way too much work, takes forever, costs are insane and… tl;dr not feasible for a bootstrapping project. Imho.
- However, when someone finds your article and requests a narration in individual capacity — that is not considered distribution (tweet me if I’m wrong). It’s like hiring a personal voice over actor, whom you pay to narrate your texts for you. This could be an option…
You can read all the drama and conflicting views here.
In the end of the day, I did not see this legal issue as signal to stop. You can find a gazillion legal issues in any business idea, when you throw several lawyers, big $$$ and angry competition in the mix. So I continued validating ReadByHumans.
Key lesson learned to HN growth — have a meaningful conversation, not just dummy upvotes. I think I did that, more or less, and we were on the first page within a few hours. Got us more traffic than PH. 🤔
Results. Let’s measure!
After about a day, it was clear that I was getting about 22% conversion rate. From landing page views to email opt-ins.
You can’t really read it on the screenshot above, the way I derived with 22%, is by dividing roughly 1270 visitors by 276 signups.
Most traffic and sign ups, as you can see above, came from Hacker News. Most Signups from Product Hunt.
I’ve never really thought about funnels before this project. I drafted a list of steps that users would have to take to start paying us money, if we were to sell a Premium tier. Those steps would be:
- Land on page (Traffic)
- Sign up with email (Subscribed)
- Add private podcast feed url to their favorite podcast app and start getting FREE value from us. (Active Free)
- Eventually start paying for more value (Paying)
In the right column, titled Facts, you can see the real numbers that I’ve got so far. In Conversion % I’ve divided these numbers by each other. Except for th 2% conversion rate. There, I just used a very optimistic conversion rate of 2 percent. Typically it’s closer to 0.5–1% for consumer products.
This modeling, using facts allowed me to project amount of traffic I had to drive to get to desired number of paying customers.
E.g. I wanted 20 paying customers and then, Desired Qty column would derive Active Free, Subscribed and Traffic numbers:
20 / 2% = 1000 Active Free (target)
1000 / 29% = 3450 Subscribed (target)
3450 / 22% = 15875 Traffic (target) — unique hits on the landing page.
Very powerful thinking technique. I was blown away! I was not planning to do anything right away with these numbers, but they were setting clear expectations and goals, if I chose to grow this project.
Hold on! By the way. Knowing industry averages is critical to judging a health of any business. When I was starting out I did not know which numbers are healthy or not. Even today, some averages is a mystery to me. For example. 25% conversion rate (qty of people who open your site divided by qty of people who enter their email) on the landing page is quite good. 0.5–2% conversion rate from freemium to paid is okay for consumer products and games. 20% open rates of your email campaigns is also healthy.
Again, why is this important to know? — Because if you are new to the Internet game, you get 50% conversion rate but you expect 100 visits and 90 signups — you’ll decide that you are a loser and stop your business, while in fact you are rocking it.
Seems like ReadByHumans had good, healthy numbers. However getting those 15k hits (ideally for free) was not immediately obvious. Tweet me if you know how, please 🙏
For the upcoming several weeks, I did not want to invest more time in growing sign-ups. Instead, I wanted to focus on making my service loved by existing users. I emailed each of them personally, trying to understand why did they sign up, their expectations, their needs, and what would they pay for. Yes! It’s a good idea to ask for money ASAP. This was not meant to be a Silicon Valley type project, where I’d raise 1.5 Gazillion of Dollars and subsidize growth forever without thinking of making a profit and sell to Hooli. I wanted to validate if people would pay for something here and do that fast.
I learned a few lessons on how to get genuine feedback from users and previously covered them in my other post “You dont have to reply”
So why did we shut down?
Well, this all was not too bad. But there were few major issues.
Article copyright was a serious issue and I’m not a subject expert. Fear of things you don’t know you don’t know is real. Hacker News reaction reminded me that no one gives a f*ck about your positive intentions. I’ll say no more.
I was spending about $300 a month on voice actors. I did not have an infinite budget and wanted to monetize ASAP or shut this thing down. Sooner or later I had to send out a sales email offering to pay for Premium service. The offer was “TOP 3 articles narrated on weekly basis + 1 FREE weekly article”. Only 1 person gave me his/her credit card information . 1 is better than 0. I smiled 🙂 Time was running. I could no longer sustain spending $$$ on voice actors, do all the weekly newsletters, article selection, audio editing and uploads. I gave myself time till the end of July, to find a way to break-even or shut down. And well, I had to shut down.
Here are some calculations that I did to estimate what it takes to break even. All numbers in bold are variables, that I played with to see what happens to the Profit cell.
Lessons learned 🤓
- Stay true to yourself and build what you yourself want. Don’t waste time for indirect routes and stop looking for shortcuts.
- There are a ton of ways to validate an idea without writing a single line of code. Even when you think you should, try your best to find an off-the-shelf solution. Case in point: did not build a backend — just a landing page. Did not build mobile app — relied on existing podcast apps to deliver content via RSS feeds.
- Know your funnel! Think in terms of funnels and apply them everywhere. Including dating. I never really thought in terms of funnels before Andrey forced me into it. When you know your conversion rates, you can suddenly start planning actionable targets all the way to break-even. Example: you calculated that you need 100 paying customers at 9.99 a month to break-even. You learned your conversion rate is ~0.5% into a paying customer per month. Therefore you need to drive 20,000 quality leads to your landing page to eventually hit monthly break-even.
- Talk to your customers. Again and again. Talk to them all the time. Get their feedback and quantify it on as many decisions as you can. Idealy call them. Forget Messanger and WhatsApp.
- A popular idea and a business idea, that makes economic sense, are not the same. ReadByHumans was a popular idea. Still is.
- Markets worldwide differ. What might work in the USA, might not work in Europe or Asia. And vice versa.
- On the state of Text-To-Speech. When we sent some lower quality human narrations, our users thought it was synthesized. I thought an average human being would be able to distinguish between robotic and human voice. Wrong! The line gets more and more blurry. People also seem to have very high expectations from AI, since they confuse humans voice with a TTS output.
Thank you ❤️
Special thanks to all who helped me with ReadByHumans along the way! Close friends, Startup School batch members, people who reached out and offered feedback. Thanks Andrey for kicking my ass. Thanks, Adelina and Ritesh for reviewing my writing. Thanks to Jennifer M Choi, Dan Siegel for providing feedback in the early days. There are too many too many worthy names here, as the article was already too long! Thank you all who subscribed to ReadByHumans and chatted with me over email and skype!
Stay in Touch! ✉️
Hope this was useful! Please share this with all your Medium friends and hit that 👏 button below to spread it around even more. You can also engage me on consulting basis, if you need some js, react, node, ui/ux or growth help. Happy to help!
Originally published at ksaitor.com on September 27, 2017.