The Impact of Safari Top Sites on Your Bounce Rate
The new beta version of the Safari 4 browser is jam-packed with new features: 150 according to Apple. One of the most appealing features is called “Top Sites”, which is the ability to have your favorite sites shown as graphical previews, as shown below.
What’s so special about Top Sites? Besides the fact that it provides a nice graphical view of your favorite sites, it lets you know if they’ve also been updated since your last visit. That’s right. Safari automatically fetches for new site content in the background and lets you know (using the blue stars in the right corner) that your top sites have been updated.
While this is very convenient for the user, it introduces new challenges for web analytics practitioners. Primarily, the inclusion of your site in your visitors’ top sites means an increasing amount of artificial traffic. Primarily, if you’re in a situation where users of Safari place your site in one of their top sites and don’t end up visiting your site, the traffic will likely show as bouncing traffic. This results in an increase in your bounce rate even though the visitor never entered your site.
We have tested the inclusion of a number of sites in Top Sites and have seen the traffic registered in web analytics even though we never visited the sites.
Safari 4 is currently in beta, so only a small percentage of your audience is exposed to “Top Sites”. Once generally available, this is sure to cause an increase in bounce rates for some popular sites. More importantly, the Safari market share has been on the rise, according to the figure below by Market Share. The latest figures show the market share at 8%. This means that as Safari continues to gain traction, this problem is bound to get bigger.
So as you’re analyzing your data, keep this trend into consideration as it could impact your overall bounce rates.
During your testing was there any special about the Top Site traffic that would allow it specifically to be filtered? The other option would be to filter out all single-access traffic from Safari, which would artifically lower your bounce rate because some of that traffic would have been legitimate bounced traffic. Or would you recommend, when calculating bounce rate, that you filter out ALL Safari traffic?
Thanks for the comment Brad.
Your solution is probably the best workaround for the time being, but I’d have to imagine the folks at Apple are working on this.
In fact, the default home page for Safari is apple.com so it looks to me like the company suffering from this the most when it comes to analysis is Apple.
Same goes for Fast Dial in Firefox (and more of the same kind).
Computer World The recently-released version of Apple’s browser could have implications for your website reporting. I recently noticed a significant increase in the bounce rate on one of our client websites. As any web marketer will tell you, bounce rate is one of the key metrics of website performance. It’s a measure of single page visits to your website – people who arrive at your site, decide it’s not for them and then leave. Measuring bounce rate by browser type in Google Analytics. Intrigued See the article here: Safari 4 – how its Top Sites view can mess up your stats.
Maybe one solution would that browser that automatically fetch content on website identify themself differently (eg. Safari-AutoUpdate/5.22.11 …) So, this way Analytics and other could filter this false traffic out.
Hi Alex,
Regarding analysis, you could always look at the bounce rate per browser to see if you’re in fact getting a higher bounce rate for the new Safari browser compared to others. If so, then you know the increase is very likely due to Top Sites.
Thanks for your updates regarding the beta version.
I just started noticing adverse impact on bounce rates being reported in Google Analytics for four sites I monitor due to this. If you create an advanced segment that matches for: “Browser = Safari” AND “Java Support = No” AND (“Operating System = Macintosh” OR “Operating System = Windows”) it catches most of these “Top Sites” page loads. Its not 100% accurate because it catches a small quantity of other legitimate traffic–but at least it gives you away to look at your reports without the distortion! I agree that they should identify these page loads with a unique user agent.
nice post i think Safari features built-in support for Apple’s Voice Over screen reader in Mac OS X. Voice Over describes aloud what appears on your screen and reads the text and links of websites. Using Voice Over….
http://www.alaxlimousine.com