I analyze web logs and I watch browser trends. I’ve watched browser usage intently for years because it has a direct impact on my testing. If I’m currently working with a web application and running a test lab, I shift browser coverage and update my test lab based on log analysis. Browser trends influence my work.
Here are a few things I learned or in some cases, saw reinforced from a browser analysis session I had this week.
(Your site stats may vary, these comments are based on an analysis I ran this week.)
Browser versions fade away over time; old browsers don’t disappear too quickly.
A close watch of IE7 before and after the holiday season showed:
A month after the holidays (February) showed a significant spike in IE7 usage. I’m guessing preinstalled browser versions on a new PC altered the usage numbers.
Two months after holidays showed a sharp decrease in IE7 and an increase in Firefox usage and an increase in IE6. My interpretation of the shift is people had difficulty adjusting to IE7 and moved to a different browser and some moved back an IE version. Something I consider with stats like this is whether the site I’m testing is functioning well with the new version. People won’t switch browsers that easily because it takes effort. So are there browser version issues in the site I’m testing that I need to track down?
Four months post holidays showed a significant decrease in IE6 and increase in IE7. A line chart with four months of data offers a more gradual view of the transition.
This is something I’ve seen over years; new browsers do not get adopted overnight. The change is gradual which makes sense. When it comes to planning browser version testing, it means there are months of overlap and it can be a long while before a browser gets dropped from testing and eventually from dropped from support.
Firefox usage is continuing to increase slowly while overall IE usage is decreasing.
This is the type of stat I begin to watch more closely. Is this a temporary change? A deeper analysis might point out different customers have brought different browser stats but I opted not to investigate at that level, I could, I chose not to. Or it might show the change to IE7 hasn’t been favored and that people who switched to Firefox may remain in Firefox. Only time and the stats will tell.
I try to maintain a separation of the actual statistical information from my interpretations because once both stats and opinions get mixed together, it can be difficult to distinguish one from the other. And it’s important to keep facts and opinions separated and to recognize one from another.
It’s also good to run an analysis and plan testing based on what the logs have to say. If you watch web logs closely, they have lots to tell.