Analytics is one of the best tools available for tracking a wide variety of metrics related to your website. Want to know how much traffic you are receiving for any given time period? Covered. Want to see how visitors are moving through your site after they reach it? You can do that too. But Analytics is only as good as the user behind the keyboard, and sometimes the data you are seeing may not be entirely accurate.
Now, there’s no need to worry, you’re probably fine, but I want to highlight some things to check for so you can be sure your data is accurate and up to date. The situations below occurred with clients of ours and after taking some time to clear them up, their data was cleaner and we knew the numbers were accurate.
Running experiments provides valuable insight into what works and what doesn’t when it comes to making updates on your website, but if they are not setup properly they can skew your regular reporting data. Below is an example from one of our clients that was running tests on their homepage – and at one point had three different variations being tested.
As you can see, as traffic to one version decreased, others were gaining traffic and ultimately it was being shuffled around to the different versions depending on the state of the test. To put some numbers behind these, for Q2 vs. Q1 the original homepage had a loss of 1.30% (16,145 vs. 16,358). However, for that same time period, another version of the homepage had an increase of 5,411.81% (7,937 vs. 144)! This more than made up for the decrease that the original homepage was showing. While all of this data can still be tracked and attributed to the homepage, seeing unsightly traffic decreases pop up when they aren’t expected can be prevented with a few quick steps before and after setting up experiments.
When creating an experiment for the first time, there is a setting titled “Consolidate for Content Reports” under Advanced Options. By changing this to Yes, Google will aggregate the traffic to the variations pages for general reporting in Analytics and track it as the original page.
In the example of the home page described above, by allowing this feature, all visits to www.domain.com/index.php and www.domain.com/index2.php will be tracked as a visit to www.domain.com/ (the original page). This will remedy the issue of drastic traffic decreases, but will not affect the results being tracked in the experiment. Easy right?
Another thing to remember is that after an experiment is completed, the Analytics Experiment code needs to be removed from the affected URL. If it is left on the page, the URL still includes the experiment ID and other related data, rather than the clean URL. While this isn’t generally an issue, it can present some concerns down the road especially in regards to backlinks. For example, if a user wants to link back to the pricing page, but there is still an experiment id in the URL, then the backlink will include the full experiment URL when the visitor copies the URL from their address bar.
Clean URL: http://www.domain.com/products/books/
Experiment URL: http://www.domain.com/products/books/?utm_expid=3535964-34&utm_referrer=http%3A%2F%2Fwww.domain.com%2Fbooks
Removing the experiment code after the experiment is completed will ensure that the URL goes back to its original unaltered state.
Websites are constantly evolving with updated pages, URLs, content and many other things. However, if a page is removed, and you are comparing long term trends, sometimes that old page sneaks into the data and causes some discrepancies. Take the screenshot below for example.
For the same time period in 2012 vs. 2013 1,069 visits were lost to this conference page. What happened?
Well, after visiting that page in the browser, it turns out that this particular client had moved that page to a different domain. The conference portion of their site reached a certain depth and it made more sense to give it its own URL. Because it was now irrelevant to our tracking purposes on the original domain, we could filter it out and it would not be included in our total traffic comparison numbers.
Finding these missing pages can be a bit time consuming, but it will clean up data and make sure you are tracking relevant pages that are currently live on the site.Also you can search online for a good carbonite offer code and use that. One method to find them is scrolling through the rows of pages in Analytics and looking for large gaps in visits. An easier way is to pull that list of URLs from Analytics and run it through a crawler to look for any 301 or 404 status codes (click Export at the top in Analytics and choose your preferred format). After you’ve identified any pages causing errors, you can dive a little deeper and make decisions on whether you want to exclude them or not.
Finding discrepancies in your Analytics data can be tedious, but by following these steps you can minimize the errors and ensure that your data is clean so you can rely on it for confidently analyzing your website’s traffic.