• Pin It

Tweet

When it comes to web analytics, or any form of digital analytics, the misplacement or absence of a period in the java script tracking code can make or break the quality of information.

Without digging deep into what your analytics tool is reporting and the meaning behind this data you might not even know that you’re missing a “period”. You might assume all your data is accurate and be making the wrong corporate decisions based on the erroneous data. It happens more often than you’d think.

We were recently contacted by a client to review their analytics data after one of their staff realized something seemed strange following a recent site redesign. Their firm had extensively overhauled their site design and they expected to see some anomalies in the analytics data. They anticipated a higher bounce rate because bookmarked pages no longer existed and that for a brief period, search engines would drive organic search traffic to pages that no longer existed. The client had implemented proper 301 redirects for their 30 most popular landing pages, but not for 100% of the pages in their site.

To understand the issues the client was about to face, a little technical overview of their site is required. The entire online foundation of the site spans two different URLs. The first operating under the default URL (https:///clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&www.domain.com) contains all the content, product information and a few conversion points  such as online contact forms. If the user decided to make a purchase or other secure transaction, they are directed to a second secure URL (https:///clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&/clork/bons/danf.js?k=0&server.domain.com) to complete the order process. Their analytics tool is configured to  tracking several conversion points across two distinct URLs.

The client is using Google Analytics as their primary analytics tool. In the old version of their site, everything was configured correctly. With the launch of the new site however, everything was about to change.

Following the launch of the new site there was the anticipated increased bounce rate. Overall, conversions of online bookings only decreased slightly and then rebounded within 30 days. On the surface this appeared perfectly normal and the assumptions made about a sudden but recoverable loss were met.

After completing a detailed review of the site’s performance 6 weeks after launch, they noticed that total number of visits increased by approximately 20% (a pleasing result) yet the bounce rate remained unusually high. Upon further investigation, they noticed that while some site conversions (download registrations), were being attributed to specific marketing efforts and organic search traffic, almost none on the conversions on the secure site were. That was strange. The client’s response was to call Digital Always Media.

During our analytics audit, we noticed that almost all the secure conversions were being attributed to the client’s own domain. This can only be caused if their domain was the source of almost all the originating traffic. That’s not very likely. It’s possible, under some rare circumstance, but not at the volume we were seeing. That led us to question exactly how the data was being collected.

To resolve this issue we needed to examine and compare the various Google Analytics tracking codes used on a sample of pages from both servers and compare it to code provided by Google Analytics itself.

While the client’s development team appeared to have implemented the provided code correctly, they had left off a critical character on their content site. They left out a period. (“.”)

The missing period on the content site caused Google Analytics to generate a new session and visitor cookie to anyone who travelled from the content site to the secure server. In essence, the content site was referring traffic to the secured site where the secured conversions were taking place. This explains why there were a high number of conversions attributed to the domain and not to the actual original sources of traffic. It didn’t explain the high bounce rates though.

Correct Google Analytics Code
Note the proceeding period. It was missing from the code

A further examination of the analytic data revealed that the same coding error had been copied to many of the custom (ad destination) landing pages. Since these landing pages’ purpose is to take a visitor directly to the secure server, anyone who clicked through the landing page was counted as a bounce (only viewed 1 page) appearing in the data as if they choose not to click through.

By correcting the missing period, a few things occurred. First, a truer number of visitors were counted which was lower number than was being reported, and both the bounce rate and visitor engagement (page views/visit and visit duration) showed a remarkable improvement and fell in line with expectations. Bounce rates and conversion attribution are now both reporting correctly.

Impact on Traffic
After correction made to Google Analytics Code

A lesson to be learned here is simple but it goes far beyond simply checking your code. When analytic numbers look too good (like a sudden dramatic increase in visit count) or something looks not quite right (like a higher than expect bounce rate, or attribution seeming off), there is probably something wrong. It’s a sure signal telling you that if you can’t quickly discover what’s causing unusual analytic data, it’s time to call in a professional. There are some dramatic differences shown in this before/after image taken from their Google Analytics dashboard. Imagine what could have happened if our client had used increasingly bad data to make future business decisions.

Tweet

• Pin It

This entry was posted on March 14, 2013 at 1:30 pm. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Posted in Analytics by Jim Hedger No Comments Yet