Improving your applications by statistics
If you developed an application and published it in a website, you can use things like google-analytics data to adjust your application. E.g. If everybody is accessing your website with 32-bits, 24-bits and 16-bits screen resolutions, you can discard all considerations about 8-bits modes in your application.
Things you should add to your application:
- A way to get usage data from your users. E.g. where they do click, how they call commands (menus, keyboard, toolbars, ribbons, etc.). In this way you could focus improvement in the more common methods to run commands.
- A way to count how many times a command was used. Maybe there are commands that are rarely used, these commands could be more deeper in menus hierarchies and without keyboard shortcut.
- A way to known systems where your application is running. E.g. operating system, version, screen resolution, etc. Improve usability in the most used platform.
- A way to known how many computers are running a specific version of your application. If a version is not used anymore, you can stop supporting it (E.g. Adding patches for a version that nobody uses).
All this must be done carefully, with proper check-boxes like “Do you want to send anonymous data about your usage to improve customer experience?” in installers, splash screens, etc.
There are some frameworks to do this kind of work. Anyway you basically need:
- An HTTP server to accept packages from users (e.g. through POST requests). Each request should be processed by a script (in PHP, ASP, etc.) which should convert the received package into database model.
- A way to differentiate users (you could SHA1 some string like “local user name + pc host name + app name version”). Here I don’t know if the IP could be a good way idea (NAT problems, etc.).
- A way to make HTTP requests from your application (firewall + proxy problems) of all collected usage data of the user (identified by the SHA1).
- A mini-program to process all the database and print some useful reports.