How did you build it?
We knew it’s important to validate our idea before jumping into coding. We followed the “The Lean Product Playbook”, we create personas with their struggles and needs.
We work to refine the feature list that we will build to our MVP, we also created a landing page and sent cold emails to professional streamers. We got few responses from streamers who recognize the problem from their daily routine, and a lot of silence from others.
Taking all we learn from that week, we jumped right into coding, using Computer Vision algorithms and OCR to detect objects and text on the video, setting up our environment on Amazon servers and Lambda functions.
We worked on it for three months, writing code and nothing else. We loved it! We had no knowledge in Computer Vision so we needed to learn it from scratch. We use Node.js and found a great open source library for Computer Vision, Opencv4nodejs, the owner of the library was very helpful and kind, thanks Vincent Mühler. He helped us get started and after a few weeks, we were able to analyze a video stream from different perspectives. First, we tried to figure out what is currently happening in the game so we started to look for text messages that were displayed in the game such as “Victory Royale” in Fortnite, this way we knew that we have detected an interesting message, the streamer won the game.
Another perspective was looking at the messages frequency at Twitch.tv chat, we figured out that when there is an anomaly in the average messages per second, something interesting might have happened and the viewers are going crazy.
So by combining those and some other “secret” indicators, we have figured out pretty well what are the interesting moments and extract them.
We still had a long way ahead of us, but we minimize editing time of a stream from ~12 hours to ~4 hours!
Three months in, we wanted to get some beta testers, but we failed to make connections in the streamers world. The streamers who were willing to talk to us before didn’t answer our emails now, and it was hard to find professional streamers to try our product in other social networks. It was our first obvious failure.
Which were your marketing strategies to grow your business?
We failed to understand how important it is to have marketing strategies and build an audience around Twitch Highlights.
Yes, we tried sending cold emails, post on forums and Reddit. But it’s hard to get anyone attention this way for a brand new product when no one knows you.
Which were the causes of Twitch Highlights failure?
We failed to build an audience around Twitch Highlights. After we couldn’t find any beta testers it’s was hard to continue, our motivation and momentum were declining fast, very fast! There was no one to give us feedback on our work and tell us it was useful for them. If you can’t help others with your product you have no energy to work on it.
Instead of confronting with our real problem, we tried to improve Twitch Highlights by adding more features, classic mistake. It didn’t took long until we realized we fail to execute on our idea, we run out of money and with no potential users on the horizon we had to go back to work.
Which were your expenses? Did you achieve some revenue? At the end, how much money did you lose?
Some numbers, we live in Israel so we were burning about 2500$ a month from our savings just to put food on our table and rent an apartment (We are both married).
We spend about $500 on AWS servers and lambda. And we worked on it for about 8 months.
If you had to start over, what would you do differently?
Tzelon:
I definitely would work very hard on building an audience around Twitch Highlights, and my personal online presence before I quit my job. I always thought that be 100% on something will guarantee its success, but I now think we should as makers, first build a platform on which we could build our products, and building this platform takes time, months even years, it includes knowing who we want to serve, get involved with that community, building our reputation with them and really live with them. I'm here to run a marathon, not a sprint.
Ron:
I would have spent more time on networking and building a community around the problem first, before going after the solution. We have started working on the product without gathering enough “intel” from our audience and actually make sure that it is a real pain for them and they do see the value of our solution.
Because we are developers the easiest part for us is the actual execution so we just jumped into it as fast as we can, even though we knew that the validation and market research part is crucial.
So I would definitely spend more time on research, planning, validating before writing the first line of code.
But I think that the main reason that we rushed into building the product is that we were scared that our time will run out, in terms of our savings.
Which are your favorite entrepreneurial resources?
Tzelon:
I love Justin Jackson (@mijustin) podcasts, “Build you SaaS” and “Product People” and his blog.
Great books I recommend for every entrepreneur are “Getting Real” and “Zero to One”.
There are great communities I’m part of like IndieHackers and MegaMaker.
Ron:
I consume most of my knowledge from audio, mostly podcasts and audiobooks my favorite podcast is “Hashavua”, a podcast in Hebrew that was made by two incredible guys that shared their experience and interviewed so many great entrepreneurs, I have listened to the entire show twice :), very recommended for Hebrew speakers.
“Lean Startup” is also an amazing resource and can really help in working in the right process.
Where can we go to learn more?
Tzelon:
You can follow me on Twitter, I’m now working on my personal blog.
Ron:
My Twitter handle is @ronco1st.
