Compendium Blogware's corporate blogging software has come a long way in a very short time. We've been adding new features and making changes to our application at a very rapid pace, mostly based on Ali Sales' and Chris Baggott's excellent knowledge of our market with a healthy dose of input from our support folks. That works fine for a startup company, but as a company grows it needs to get more scientific in its approach. Decisions eventually need to be made based on hard data.
The trick is to find a way to know for sure that a change you're making to your application is a beneficial one, and that information can only come from experimentation. Three men from Microsoft's Experimentation Platform Team have written a paper that describes how such experiments are performed.
The basic idea is that the new feature is shown to only half of your application's users for a given period of time, typically a week or two. The data collected during that time will tell you how effective the change is. Did the group of users who got the new green buttons on their interface complete their work faster and more accurately than the users of the old yellow buttons? The data will tell you.
User behavior can change dramatically as a result of even very small application changes, and often in very unintuitive ways. Get the real data and get rid of the guesswork.
Practical Guide to Controlled Experiments on the Web: Listen to Your Customers not to the HiPPO: http://exp-platform.com/hippo.aspx
The trick is to find a way to know for sure that a change you're making to your application is a beneficial one, and that information can only come from experimentation. Three men from Microsoft's Experimentation Platform Team have written a paper that describes how such experiments are performed.
The basic idea is that the new feature is shown to only half of your application's users for a given period of time, typically a week or two. The data collected during that time will tell you how effective the change is. Did the group of users who got the new green buttons on their interface complete their work faster and more accurately than the users of the old yellow buttons? The data will tell you.
User behavior can change dramatically as a result of even very small application changes, and often in very unintuitive ways. Get the real data and get rid of the guesswork.
Practical Guide to Controlled Experiments on the Web: Listen to Your Customers not to the HiPPO: http://exp-platform.com/hippo.aspx
Comments for Better Web Applications Through Science
Leave a comment