Recently we converted our Web analytics over from NetTracker, a log-file based analytics tool, to Google Analytics. After the initial vertigo that accompanies the jaw- and traffic-dropping shift from log file to page tagging analytics software I began to look closely at some of the metrics that Google Analytics offers.
As media companies, we tend to focus, with some bravado, on our traffic--this many visitors, that many visits, and this many pageviews.
But the fact is, many times visitors come to the site, realize that they've hit the wrong site, and leave. This is called a bounce. I'm pretty sure most publishers aren't reporting this metric to advertisers.
The bounce rate is basically the number of single-page visits as a percentage of all your visits. The intent behind the bounce rate is very good. It asks us to focus on visit engagement or quality, and shift away from the raw numbers. Put another way, if you have a site with 100K unique visitors per month, but 88% of them came once and left, is that such a good thing? Is that audience really engaged?
When content site owners first look at bounce rates, they can seem rather scarily high. Across our sites, bounce rates range from 50% to 80%.
When I shared these with our sales team at a recent meeting, there was mild panic in the room. Our most numbers-oriented, analytical salesperson asked me for a set of reports that excluded those who bounced. Which would leave a puny core of die-hard users.
So, does everyone who bounced (that is, visited one page and left) count as a useless visit? No, and that's the problem with bounce rate.
Bounce rate is a good idea that's an unreliable metric for content sites. Consider these three scenarios that are typical of the way people consume content on the Web these days:
- A visitor clicks through on one of your headlines in an email newsletter to read the full story on the Web site.
- Another visitors does a search in Google, finds one of your videos in the search results, clicks through to watch the entire video (highly engaged), and then leaves.
- A third visitor clicks on an article to your site that was forwarded by a friend or colleague.
In all three cases above, I would argue those are highly engaged experiences. But Google says they're bounces. Hence the problem with bounce rates.
Bounce rates make sense for e-commerce sites, which of course, is what drives a lot of Google's advertising business, and hence, its entire product suite--a relentless focus on conversion optimization. Bounce rate figures into that.
But for content sites, bounce rate cannot be considered a reliable measure of engagement. What is?
Return visits and time on site. The're oldies but goodies. (For more, I recommend reading this article on audience metrics on the newly launched eMediaVitals web site, which is the most intelligent site I've seen on emedia, hands-down.)
A new metric is emotional reaction. How do visitors feel about our site? Maybe 100K people visit the site, but 90% of them hate it. That doesn't get captured in any traditional analytics software. That's why we've started using Kampyle on our sites--it's a brilliant way to unobtrusively capture feedback from site visitors, with a little device that hovers quietly in the corner of every page. Definitely one of those "Why didn't I think of that" ideas.
And of course, my favorite metric, outcomes. Some folks call them conversions. I.e., how many people signed up for a newsletter or registered for a white paper or video? We recently learned that for one of our brands, we had over 80,000 registered users register for 290,000 pieces of content over the last three years. That metric didn't even come out of Google Analytics. (It came from our LeadWise-powered AccelaWorks platform.) And yet it is the most accurate measure of engagement that I can think of, because it's where the rubber meets the road -- a web site visitors taking a defined action that creates value for both them and the advertiser.