Benedikt Foit, Cornelius Rabsch and I founded Tagcrumbs in the summer of 2008. We were aiming to create a platform that could be used on the web and on mobile phones to share and remember your favorite places. We were not only focusing on reviews of restaurants or businesses, we also wanted to provide a way to remember the best place to have a picnic, the place of your first kiss or the best place for kite surfing.
We worked on the platform for about 1.5 years; in the end all of us were working on it full time. We decided to shut Tagcrumbs down in the beginning of 2010. You can find some reasons on my blog post. I was responsible for building the infrastructure, the underlying platform (without the front-end) and the iPhone application of the service.
The Platform
We built the service on top of Amazon EC2 and EBS to allow easy scaling. Postgres and the Postgis extension were used as a geospatial database to store all the places and to allow fast geo-queries. Ruby on Rails was our framework of choice, Nginx, Mongrel (later Passenger) and HAProxy were used to serve the websites, Monit for guaranteeing uptime of services, Postfix for sending all the emails and Cacti for graphing resource usage levels and application statistics.
The RESTful Web Service
I created the RESTful web service that was used as the foundation for our own iPhone application and Tags2Go, the Android App that was built by our partner Sapient. The API was using OAuth for authentication and was available in a unified XML and JSON format. Our own DSL was used to provide several different data representations without any hassle. The API followed many more RESTful conventions (e.g. URLs as identifiers, links to other resources) than the standard APIs provided by Rails. An extensive documentation allowed our partners to built applications on top of Tagcrumbs.
Tagcrumbs iPhone Application
We wanted to build our own iPhone application to show what’s possible with our platform, so I taught myself how to use the Cocoa Touch frameworks and program in Objective-C. The app was built on top of the amazing Three20 framework by Facebook. Three20 was difficult in the beginning as there was hardly any documentation, but I learned plenty by diving into the code, and in the end it allowed us to built a much better application.
The application was featured by Apple in the German App Store and we received a “Mobile Talent Award” for it by the German government.
Our Achievements
Tagcrumbs was covered by quite a lot of blogs, newspapers and magazines (Gadling, TechCrunch Europe, Deutsche Startups, Internet World Business, NTV, Netzwertig, Mobile Zeitgeist, and many more). And as I just mentioned, we received an award for our iPhone application and got covered by Apple in the German App Store.
Shortcomings & Learnings
In the beginning we planned with an ad-based revenue model. We thought mobile and local ads could allow us to earn enough money to support the business. This was not the case and we never reached enough scale to become profitable with regular ads. We thought about many different ways to solve the problem, one was offering services around Location Based-Services to businesses, but in my opinion we decided to halfheartedly to implement them. We always wanted to keep the community and platform alive. We should have pivoted hard – and left the community behind – as soon as we found out that the ad-based revenue model was not be sustainable.
Trying to do a split, with only 3 employees, between the platform and business services, was probably also the main reason that caused many potential investors to run away. We were therefore not able to raise enough money to support our plan.
In the end, I don’t regret doing anything looking back. I was deciding between going back to university for an MBA or another study program and decided to do the startup instead. I think I learned much more doing the startup, got to know many interesting people and I probably even spent a lot less money.
