It's a simple fact of life: like George Lucas's continual "improvements" to the Star Wars trilogy, your perfect, flawlessly executed Google Analytics implementation will
suffer at the hands of routine website maintenance and updates. New pages will appear out of the ether, untagged and untracked. Your exquisitely customized tracking code will inexplicably have a link to a dancing hamster gif inserted directly in the middle of it. That guy in IT will somehow gain admin access to the primary GA account and make "creative adjustments" to your custom reports, citing mysterious and illusive "corporate reporting standards". Then he spills a Mountain Dew on your keyboard.
"Revenue? Why are we reporting on THAT?"
Simply put, the more people touching your site, your data, your analytics kingdom, the more watchful you have to be. The best defense is constant, proactive vigilance, but even then, things will slip by you. Designers, programmers, and developers have their jobs to do, and analytics tracking quality is usually an afterthought. A well-developed QA process and management's emphasis on its importance will do wonders here, but in the meantime, here are 3 relatively quick ways to audit your Google Analytics implementation.1. Are all pages tagged and passing information correctly?
Sounds obvious, but lots of questions on missing, incomplete, or "weird" data in Google Analytics can be answered by identifying one or two untagged pages. Without the ability to capture user data set within the GA cookie on every page, you're going to start seeing a lot of odd data in your reports. In fact, I'll go on record as saying that if you haven't tagged every page on your site, you can not produce sound business recommendations from your analytics solution.
This seems obvious, but depending on the site, a simple, untagged "About Us" page can throw a wrench into the whole works.
You have a couple different options here.
2. Clean your filters!
- Scanning tools. For sites with more than 3 pages, the above option isn't really feasible. A great free tool is EpikOne's SiteScan, which will scan the first 100 pages of your submitted domain, then generate a very helpful CSV file with the results. Premium users get some additional features, but for smaller sites, the free service typically fits the bill. For an in-house solution, WASP (originally developed by Stephane Hamel of immeria) is an extremely robust extension for Firefox that will manually crawl your site and deliver all sorts of information on a wide variety of analytics solutions, not just GA. Extremely helpful is the fact that WASP doesn't simply look at the page's source code for tracking code, but actually tracks if the code fires. The free version is limited to 20 pages per crawl, but this is such a useful tool that you'll probably want to purchase a license if you're serious about tag audits.
Functioning as a data bottleneck or net for which information gets displayed in your reports, filters in Google Analytics
offer a fairly advanced level of customization. However, improperly formatted filters can result in inaccurate statistics populating your reports - or, in fact, block important data from being passed to GA at all. When first setting up a new filter, or if multiple users have access to edit existing filters, make sure to:
3. Verify your incoming link tagging structure.
- Always maintain a default, unfiltered profile. Since filters function as a kind of "data net" that prevent certain information from showing up in your reports, it's important to realize that once you apply a filter, that data is gone. If you set up an exclude filter on February 1st to block all internal traffic from appearing in your reports, then realize on the 28th that you've somehow managed to exclude all tracking data on visitors coming from Google, well...that's that. That data is gone, with no way to retroactively find it, and now you've got to explain to your boss why organic search traffic dropped 95% in February (and why you didn't catch it until the last day of the month!). Always maintain an unfiltered profile.
- Brush up on that RegEx. Regular expressions offer an extremely flexible way to craft your filters, but can be fairly confusing at first. An inadvertent "." can completely change the data set you're capturing, so be sure to take some lessons and make use of another handy EpikOne tool, their RegEx tester. If you're simply trying to exclude internal traffic from a certain IP address, or a range of IPs, you can verify your coding skills with Google's RegEx generator for IP addresses.
If you have control over an inbound link (say, your specified destination URL for a Bing PPC ad), you can append the URL with some tracking variables to properly categorize that traffic in your reports. For example, my Bing ad, when clicked on, might take visitors to example.com/product.html?utm_source=bing&utm_medium=cpc&utm_campaign=adcampaign1
- the bolded variables in this URL are captured by Google Analytics and let you know that the visit came from Bing, but was in fact from a CPC (cost-per-click) ad within your "ad campaign #1". Now that this information is tracked, you can analyze these visits in the same CPC reports you use for advertising in Google and do some nifty slicing and dicing.
I've covered link tagging in Google Analytics
in more detail before, so here I'll just say - make sure your variables are logical
Don't tag paid Bing traffic as "ppc" when Google AdWords is grouped under "cpc". Remember that these tags are case-sensitive - "keyword" and "Keyword" will
show up as two different items in GA. If you're doing banner advertising and paying for bundles of impressions, not clicks, don't throw that in with your pay-per-click campaigns! Doing so will only make analysis more difficult on your part - a well organized hierarchy of link variables can make the difference between spotting a staggeringly awesome opportunity and blowing past it with a "Traffic Sources" report that runs for 3 pages.
A lot of issues with Google Analytics revolve around these 3 items, or a combination thereof, so it's important to audit regularly
A corporate push for data quality and a rock-solid QA process are incredibly helpful, but a weekly rundown of these 3 areas (using some of the tools above) can dramatically reduce lost data, identify potentially inaccurate statistics, and hopefully pinpoint some opportunities or areas for improvement!
Labels: Google Analytics and Website Tracking