- Posted by liammclennan on September 19, 2009
UPDATE: Since writing this post shortestroute.info has been renamed to TheFastWay.net
I always like to have a side project. Something I work on in my own time, as a way to sharpen my saw and hopefully build something useful. In May I wrote about what was then my new project. To build a web site, and possibly a web service, that solves the travelling salesman problem. Now that the site is live I am writing this post to document what the process has been like.
Having a long history of building things that no one wants I decided to test the market a little. I setup an information site at http://shortestroute.info with a form to collect email addresses and a google adwords campaign. This was not my first adwords campaign but it was by far my most successful. Adwords currently drives 6,000 visits per month to the site. Here is what my analytics stats look like over the period.
Of the 17,360 visitors to the site 4% (702) of them registered their interest by completing the form. When I get around to it I will extract the 702 email addresses and names from gmail and send them an update (anyone know how to do this?). Of course, adwords is not free. Since the campaign began at the end of May it has cost me AU$1,347.14; close to $2 for every person who completed the registration form. The average cost for each of 19,666 clicks was AU$0.07. Those clicks were generated from 545,642 impressions. Given that the application is providing a free service this business model is clearly nonsensical. However, adwords provided an assurance that someone thought the service would be worthwhile, which gave me the motivation to see it through.
Development has taken approximately four months of part-time work. 95% of the time was spent on the algorithm to determine the shortest path between a set of points. It turned out to be more difficult than I had anticipated. Calculating the shortest path by brute force was not an option because, as the wikipedia page points out, there are 43 589 145 600 possible paths between Germany’s 15 largest cities. So, my algorithm is an approximation, but so far it seems to be a good one.
From here I have a number of features that I would like to add. I also need to work on the marketing. I need to come up with a name (suggestions welcome), a logo and a new domain. The design also needs work.
Thanks to all those who helped me refine my idea, and to those who are helping with testing, especially Tarn. If you have any feedback please leave a comment or send me an email.
