Startup lesson learned: Work vertically before you work horizontally
Startup lesson learned: Work vertically before you work horizontally
In this blog post Iâm going to share with you an important lesson that Iâve learned about startups. Iâm happy to have learned this lesson in the beginning of my career as an entrepreneur, about a decade ago, because I got to apply it in all the projects that came afterwards, both my personal projects and my client projects.
The year was 2006. My friend and I have just founded our first startup, Bintos. We were both 20 years old. Weâve been good friends since high-school, and weâve always dreamed of starting our own startup and making it big. Now that we were both âadultsâ, it was only natural that weâll start a startup and that weâll do it together.
Bintos did something similar to Khan Academy, before the latter came to be famous. We wanted to produce high-quality video lectures for high-school students that would target the material they had to learn for the matriculation exams, and then, of course, put these video lectures free online. Please remember that this was 2006, when YouTube was just a young startup waiting to get acquired by Google, while Google Videos was still superior. Everyone knew that itâs only a matter of time until video on the web makes it big, but the quality and speed werenât close to what they are today.
How did we get the idea to produce video lectures for high-school students? We were both fresh out of high-school, and we remembered how difficult it was to find material to use to study for our exams. (There are a lot of books in the library, but students want learning materials that are laser-focused on the exam, because otherwise they might âwasteâ time learning things that arenât on the exam :)
We remembered that when we were in high-school, we were so desperate for relevant learning materials, that we were stoked to find a text-only website that was badly-formatted and almost impossible to read. The fact that such a shitty website was our best option was a sign for us that a change was necessary; so why not bring it ourselves.
Now, I had just dropped out of the Technion, a leading technical university in Israel. The Technion had a very impressive and successful project of video lectures for its students. Students watched these video lectures obsessively; sometimes to cover for missed lessons, but usually to prepare for an upcoming exams. I would personally watch it from my computer in my dorm room, but most people would watch it in the universityâs computer farms. They even had a vending machine giving out VHS tapes of these lectures! It was incredible.
When we stared Bintos we basically wanted to take the success that video lectures had in the Technion and apply it to the world of high-school students.
Now, we were young, inexperienced and poor. Starting a startup together was hard. Neither of us even had a job before. I wasnât a serious programmer back then. We had little experience in being adults and talking to people seriouslyâ An email that I would write today in 3 minutes, we would labor on for maybe 30 minutes, to make sure it had just the right mix of politeness, assertiveness, and knowing-what-the-hell-weâre-talking-about-ness. My point is that getting anything done was an inefficient struggle. We were the very definition of a scrappy startup. But still, we worked hard and felt that this was our calling in life.
Our main goal was to produce a sample video course that we could put online so we could get feedback from students, and hopefully investor attention. This meant we needed (a) a teacher to give the lectures, who should be as good of a teacher as we can get, (b) video equipment and knowing how to use it and Š to edit the videos and putting them online on a website.
After months of hard work, we got everything we needed. We got an amazing math teacher, who worked in a top private school, to volunteer to give the lectures. (Iâm still amazed that my cofounder convinced him to spend a few days off from work with us for no pay.) We got the private school mentioned above to allow us to use an empty classroom to film the lectures. We hired a professional cameraman, who had a high-end camcorder that produced great video; he was also in charge of our sound recording. And we pooled our money to buy another, smaller camcorder, for the wide shot, so that we could combine footage from both cameras when we edit the video.
We scheduled a few days of filming, and everything worked out great. This was the culmination of all our hard work, and we were very excited. We both sat in the back while the teacher gave the lectures and the cameraman followed him with the camera, and made sure everything went according to plan. The teacher did a great job and gave a great lecture.
After we finished all the filming, I sat down to do the arduous work of editing all that footage down to consumable video lectures. (I picked up Adobe Premiere for this task.) We converted all the footage from MiniDV tapes to files, and I took a look at the footage. The footage from the auxiliary camera was good; the sound quality was shitty because it was using the on-camera microphone, but no worries, the main camera was the one connected to the collar microphone that the teacher was wearing, so weâll use that.
I load up a video file from the main camera and play it. The video quality is great, audio is loud and clear with no background noises, everything looks perfectâŚ
A loud, unpleasant noise. I was confused. A few more seconds of perfect audio went by, and then again: Swoosh swoosh swoosh swoosh.
That noise turned out to be the collar microphone, that wasnât attached firmly enough, rubbing against the teacherâs sweater.
I looked at different points in the video file, and at other video files from the same camera. They all had those sounds all over the place. At most it would be gone for 15 seconds, and then it would come back again.
I showed this to my friend, and we were devastated. That noise was so loud and distracting, we couldnât release a lecture that had it. We considered what we could do. We tried to digitally remove it, but I doubt even professionals could do that, certainly not us. We considered using audio from the second camera, but the quality was bad, the teacherâs voice was muffled and there was lots of echo.
We ended up having to refilm the whole thing. We were very lucky that the teacher was patient with us and volunteered a couple more days to help us. The cameraman took responsibility (after some arguingâŚ), since he was supposed to be in charge of the sound, and he gave us those extra days for free. So we got to produce the lectures at the end and do it right: We made 100% sure there werenât any noises when we filmed, and the lectures came out great after I edited them.
But after the whole thing was done, I stopped to think: What can I learn from this? How can I prevent something like this from happening again in the future? Whatâs the mistake I made?
My mistake was working horizontally before working vertically.
There were several âlayersâ of work. The first layer was filming the lectures. The second layer was editing them. The third layer was uploading them and making sure they looked good on the site.
Every layer was a lot of work, and the most satisfying way to do work like that is to dive into the first layer, focus exclusively on it, finish it, and then move on to the next. I think that we like this method best because you get a feeling of accomplishment when youâre done with a certain type of tasks. Also, itâs probably more efficient because you get to concentrate better on each task. For example, if you were cleaning your house, you wouldnât wash half the dishes in the sink, then clean half the floor, then scrub half the toilet, and then go back to the dishes again. Youâd finish every layer of work before moving on to the next.
But this method (which I call working horizontally) is good only when youâre very familiar with the work and have high confidence that nothing will go wrong. (Like housework.) When youâre doing something youâve never done before, the right thing is to first work vertically, taking a sample amount of work and driving it through all the layers. Itâs less efficient, but the reason itâs better is because you learn what every layer really looks like, and you get to make mistakes earlier rather than later, so you could apply the lessons youâve learned when you do the bulk of the work.
In the video lectures example, the right thing to do would be to film a single lecture, edit it and upload it. If that came out alright, then we should have gone forward with the course. In fact, I could have probably done the first lecture while they were still filming the other ones, so it wouldnât have even wasted any time.
Ever since Iâve learned this lesson, Iâve been applying it to every project that I do.
When a client comes to me with a job and I look at all the work we have, I always insist on taking one sample unit of work and driving it through all the layers, just to catch any mistakes early. After thatâs done, then itâs time to do the bulk of the work.