I’ve spent the past week in Boulder at the TechStars office, and had a chance to meet with a bunch of great startups. It got me thinking about my failed startup, Coursevote. It’s hard to believe that it has only been a year since I started it. I never really detailed what happened, so I thought this might be a good time to share my story.
The IdeaIn college, you have many friends who have taken many courses, but there’s no easy way to know WHICH friends took WHICH courses, and what they thought of them. It would be great to know which 10 of your friends took GEOL 101 and what they thought of it, but you don’t want to ask all your friends if they’ve taken this course – most of them haven’t. Coursevote was a way to see what your friends, and everyone else, thought about courses.
There’s many services out there that tell you what everyone thinks of a course (RateMyProfessor, PickAProf…), but your friends’ opinions matter more than the crowd. Coursevote would let you see that 40 people liked this course, 5 of them were your friends, list those friends and make it easy to ask them about it.
The original monetization plan was allowing students to sell ads for used books alongside relevant course pages. So on the MGMT 211 page, students could advertise their used books for this class. I quickly gave up on micropayments from the masses and shifted to reaching critical mass at universities so I could sell ads to local businesses. It never got big enough make ad sales worthwhile.
How it got startedAround March of 2006 I decided I wanted to pursue this idea after thinking about it for a few months. I entered a business competition called the Texas A&M Ideas Challenge, and sold a content site I managed to raise some money. In May, I won and the next day I hired a developer to build it.
We worked over the summer to build the beta site using the Facebook API, which is where we got the friend connections. Right as we were testing out the beta the Facebook Platform was launched, and I decided we had to rebuild it as a Facebook application.
What went wrongI did a lot of things wrong, but the two biggest mistakes were jumping on a new, unproven piece of technology and not hiring my developers correctly.
Mistake #1 – Facebook Platform
The original plan had been to build our own website ( www.coursevote.com ) and use the Facebook API to figure out the social graph. Moving our application inside of Facebook right as the Platform was launching was terrible because the Platform didn’t work. There was no documentation, which made my developer’s life hell. And they liked to change things without notice, so my developer spent countless hours trying to figure out what Facebook decided to “fix” that broke our app.
The worst part was the constant downtime of the Platform. After about a month of talking to professors and staff of Mays Business School, I was allowed to show Coursevote at a required meeting of about 500 students. I gave the demo on a Tuesday and Thursday – the entire Platform was down Monday to Saturday that week, so all the students I showed it to went home and found my application broken.
I don’t want to blame Facebook for it all. Even if the Platform worked, moving my application inside of Facebook was a bad idea and I should have realized it sooner.
Facebook applications were mostly games, and functional applications got drowned out in the noise. The applications inside Facebook should be things you use often; Coursevote was something you’d use once a semester. And many students didn’t like the concept of applications, so would refuse to try it out regardless of its utility.
If I had kept it on its own website, I could have used the Facebook API without having to get caught up in the “useless Facebook app” wave. It could have developed its own brand, and students could have associated Coursevote.com with “where I go for course information each semester.”
The Platform added no new value, only more complexity. It transferred control of my application from me to Facebook. If I hadn’t been distracted by the shiny new technology, Coursevote might have had a better chance of succeeding.
Mistake #2 – Hiring Developers
This was the first time I had ever hired someone to do work on a startup with me. I didn’t know the proper way to compensate, and I learned a lot by seeing what didn’t work.
I hired my first developer the day after winning the business competition. He was a friend of a friend, and I had heard great things about his programming abilities. But he wasn’t passionate about the project; he was just a “hired hand” doing his job. I told him the project specs, he quoted me a price, and we got to work.
After Facebook launched the Platform, we had to rebuild almost everything. He reached the point where he just wanted this project to be done, and the code reflected that. I had to hire another developer take over.
I saw that just paying someone who cares nothing about the project didn’t work out well, so I sought out a developer who liked the idea and would work for equity – becoming a cofounder. This worked well at first, but he became less interested in the project when paying jobs came around and he saw Coursevote wasn’t taking off as well as we hoped.
If I could redo it I would have vested his ownership and paid him a small salary – letting him choose his own salary/equity balance.
Where Coursevote is nowIt's still on Facebook, but its main differentiating feature - what your friends think - still doesn't work. What's up now is Version 1, Version 2 (with the new developer) never got finished.
What I learned
- Put together a great team who loves the idea, and compensate them fairly.
- Fail early. Be willing to pull the plug when it becomes obvious it won’t work.
- Seek the advice of people who have done this before. This could have solved a lot of the developer problems.
- What your customers say and do are two different things. Everyone I asked said they would definitely use the service, but it was so difficult getting people to actually sign up for the service.
- Have more money on hand than you think you need. Nothing goes according to plan.
No comments yet