Case Study: New Visitor versus Returning Visitor Bounce Rates in Google Analytics
One of our clients had a higher than average bounce rate after a site conversion. We were initially surprised to see higher bounce numbers as the new site had a more prominent search bar, targeted conversion elements, and ranked much better for all targeted keywords.
Was the site loading slow? Did we have a design that wasn’t conversion focused? We’re we ranking for irrelevant keywords? Did we have too many links out to external sites? None of these usual high bounce rate factors was present.
Upon closer examination of the data we determined that our bounce rate was skewed upward by a coulpe of things.
1. one time visits for long tail keywords
2. returning visits (from the client employees and our SEO team)
People were finding the site for many more long tail terms since we’d switched to the new design due to the greater power of the site and the way we architected the site rebuild. We changed to pretty permalinks and grabbed a bunch of solid relevant links over the course of the campaign. We determined that though it changed our numbers some, this sort of bounce is not a problem. Google analytics data is not used by Google to determine search rankings (they have much better data than that. The Chrome web browser for example).
There were of course hundreds of these types of long tail searches tracked in analytics, but there’s really nothing to be done about these particular types of bounces.
Keep in mind, not all long tail searches bounce. Many connect and convert to the next click (another page, or the contact form).
Should we alter Google Analytics to reduce bounces or for better bounce data?
We considered changing the Google analytics script to count people staying on the site for more than say 10 seconds as a non-bounce. This would help us keep from counting people who found the product they wanted immediately and then called instead of clicking to email. We could address this by tracking the adjusted bounce rate by changing the analytics code as per:
We didn’t really think this was necessary, and decided to hold off and check a few other things first without messing up all our data. Goog to know it’s available though, and likely a smart way to track data for certain sites.
The biggest issues we found with bounce rates was with New Visitor versus Returning Visitor stats.
For example: Search term A had a 72% bounce rate for returning visitors and only a 29% bounce rate for new visitors.
New Visitors are what we care about. Returning visitors could be client in house or us running ranking reports or searches.
The rank tracker tool that we use to track keyword rankings visits a page once it finds where the term ranks to guard against our ip address being blocked by google. Google doesn’t like people loading their servers by checking rankings. I don’t blame them, and we try to be as nice as we can by using human emulation and running our reports slowly.
Addressing high search volume and high click through rate bounces
There were some high volume terms for which we ranked that were receiving high bounce rates. Terms such as these are best ignored in promotion efforts. No need to work to acquire links for terms that are not going to convert.
We only found 4 or 5 high volume search terms that had more than a dozen click throughs in the past 6 months. They all had bounce rates over 80% for new visitors and one even had a 100% bounce rate.
We of course decided to ignore those terms as terms that we promote (we identified them as a not-so-great at the start of the campaign, and hadn’t promoted them much anyhow, but due to client expectation, we had moved their placements up a bit) Instead we more heavily focused on similar but more highly targeted, longer tail terms which were lower volume but more relevant with much lower bounce rates.
Another note: The client’s domain name itself with the http:// but without www. had an 85% new visitor bounce rate. We assumed that was the client’s people or myself venturing to the site, or someone simply looking for the phone number. The client verified and said that that was their in house navigation ventured to some backend tools and resources they used regularly.
The end result of our research was that by looking at new visitor bounce rates versus previous visitor bounces, we determined that a large number of our primary keywords did quite well landing under 50% bounce rate and generally between 30-40%. We did have a number of terms between 50-75% but determined that we should still work on promoting those terms. We implemented additional category pages that were more specific and longer tail to get visitors to those pages to help reduce the bounces.
We also decided to make the search box on the site more visible and fine tuned our targeting of promotion for those terms which were converting well.