Here at Top Floor Technologies, we get to see all kinds of interesting stuff. Recently while performing a strategic review for one of our clients we almost had a heart attack. While taking a deep dive into the analytics for the client and comparing the traffic year over year, we noticed a large decrease in traffic. The decrease was about 38% in overall visits. For nearly all of our clients, this is the opposite case; so this was a cause for concern. In order to deduce what the culprit was, we need to look at the usual suspects.
Determining The Root Cause
Since we spend A LOT of time in Google Analytics and other web analytics solutions every day, we have developed a quick checklist to help figure out what might be causing this decrease. The first item to look at is to determine the exact date that the decrease started on. To do this, we take a larger view of the data. As seen in the chart below, it is quite clear when this happened. Now that we know when the decline took place, we can check our time logs to see if we did any work around that time and up to one month prior. After we looked through that and found that there was no work done on our end to the site, we need to go back to Google Analytics to further analyze where and what might be causing the issue.
Slicing And Dicing
The next step in the process is slicing the data into segments and seeing if those segments are affected to the same degree as the others. The first slice to take a look at is how the traffic changed around that date for Search, Referral, and Direct traffic. In this case, looking at those slices we see that Direct is not affected, but both Referral and Search are. It is important to note that, in this example, the Direct traffic is quite small in comparison to Search and Referral traffic. Since both Referral and Search traffic are affected to the same degree, then this tends to mean that it is not caused by a loss in rankings on the search engines. To confirm this one can check the rankings on for their major keywords. I did not go that far to deduce the problem since I noticed that not only was traffic down for referrals, but also for search traffic that came from branded terms. However, the above alone was not enough to answer all my questions. I decided to compare the date before and after the drop in traffic. As seen in the chart below, a lot of red numbers, but we knew this before. However, what we didn't know before is that the bounce rate has shot up. There was in increase of 37.66%. This is rather interesting, since if it was just a pure drop in traffic and using the law of averages, there should not be a significant change in the bounce rate.
Now that we know it is not a decrease in rankings and the numbers are not diving, it is time to check the simplest things first. The first thing to check is to make sure all of the tracking code is on all of the pages. The easiest way to do this is to go to the website and take a look at the source code. I proceeded to look at the pages and noticed that the site was taking an extremely long time to load. In order for you to look at the source, you need the entire page loaded. I poked around a few more pages and noticed that it was the same case on the other pages.
To Confirm A Hunch
There are two basic things that can cause tracking to not work properly. The first is that it is not installed, the second is that it does not load. Since the page was taking so long to load, this may be the issue and would account for all of the issues that we are seeing in Google Analytics. However, I needed to confirm my hunch. To confirm my hunch, I took a look at trusty Google Webmaster Tools. They have a handy tool in their labs section which tracks your sites loading time. Below is the resulting chart. WOW, I think my hunch is confirmed. I have never seen such a huge increase in the average time of loading a website. On an interesting side note, Google does use page speed as a factor in your rankings. But since our did not change much, it appears to have little effect.
Finding the Gas Pedal
Now that we have our problem figured out it is now time to fix it and get back to normal numbers. There are several tools out there to find what is causing the the site to take forever to load. The two tools I use are FireFox plugins, PageSpeed from Google and Yslow from Yahoo. They are both great at finding out what needs to be optimized for all things related to page loading speeds. The report below is from running a Yslow report. The one highlighted in orange is the piece of code that is causing all of the issues. The second column with the numbers are the response codes generated from the server when the request was made: 200 means it is okay, 404 means the file was not found, -1 means unknown, the server didn't respond. This basically means that the website was waiting (up to 40 seconds) for a response to load that code before it moved on to load the rest of the website. Since this particular website has the older version of Google Analytics, it is located at the bottom of the page and is after the offending line of code. That line of code has been removed and Google Analytics will start reporting accurate numbers immediately.
While your Google Analytics may be telling you that the sky is falling, you absolutely want to make sure that it is the case before you go building that bunker you always wanted. In this case, the numbers are not accurately reflecting the true visits on the site. Since it took so long to load a page, users were not sticking around long enough for the tracking code to even load and register a visit.