I spend a lot of time trying to figure out exactly what is wrong with some clients’ analytics installs. Unlike types or broken code, or misspelled words – issues with tracking scripts don’t always raise their hands and scream, “Hey dummy, you deleted the </script> tag!”
I thought it would be handy for the beginners out there to have a few tips for finding issues with your tracking and some tricks to resolving those issues without tearing your hair out.
I’m Referring To Myself…
This is probably the most common error I find, your own domain is showing as a referrer to your site. Logically, this doesn’t make sense since anything that happens on your site should be an action of course, not a referral.
Most times, this means there is a page missing a script somewhere – the question is, how do we find that page? I have a few tricks for you.
- Run a tool called AnalyticsCheckUp.com and ask it to crawl your site and give you the status of GA scripts on every page. The excel sheet you receive within 24 hours (for the free version) will show “yes” you have scripts that match your Analytics account or “no” your scripts don’t match your analytics account, or you don’t have scripts at all, for every crawled page on your site.
- Check out the “pages” section listed under content. If you don’t have a gigantic site – you might notice which page is blatantly missing – likely that one doesn’t have scripts on it, and those that land on that page, and navigate to another page are the source of your referral problems
Where’s The Money?
I’ve said it a lot, I’ll say it again, tracking ecommerce transactions is a tad more complicated than just “turning ecommerce on” in the Google dashboard.
Scripts will need to be altered to pick up transaction data. Let’s face it, every shopping cart, booking engine, reservation system has a different label for every piece of information – there’s no way to automate that in Google Analytics. You’re going to need to set your scripts to work with your setup.
Most setups are going to require you to add the _addTrans(), _addItem(), and _trackTrans snippets to your code to ensure proper collection of data.
You are going to need to configure these very carefully and test your recorded transactions against your inventory/sales management system to ensure you’re tracking every transaction one time.
For more on setting up your shopping cart to track online revenue, see the Google Code Help section on Ecommerce Tracking.
Everything Just Tanked
Undoubtedly there will be a day when you login to your analytics and the dreaded flat line will appear. The panic of “I’m going out of business” will pass, and please understand that it is likely a tracking issue as opposed to a sales issue – nothing goes from gangbusters to nothing overnight – unless you’re Goldman Sachs.
First step: make sure nobody has been messing with the site. Call IT, call design, call programming – call anyone who could have access to the code or pages on the site (even via a database only) and find out if they’ve been “experimenting” recently. It’s likely the numbers took a nosedive because the scripts weren’t installed correctly after these changes, if they were installed at all.
If nobody will fess up – run the analyticsCheckup.com tool mentioned above to make sure every page has scripts. You are likely, as in 80% or more, to find your culprit here.
We recently had a site go live with the entire Google Analytics code commented out on a section of the site. It was done for testing before going live, but someone just missed it in the steps leading up to pushing it live, and the flat line was quick and disturbing until we figured out what was wrong and fixed it.
More Bounces Than A Bouncy House
Bounce rate is one of the main indicators of guest “satisfaction” on a website. If they hate the colors, navigation, or information –they’ll look at one page and leave immediately. That being said, bounce rate is pretty easy to manipulate if your analytics aren’t set up correctly.
For example, if page 1 has a Google Analytics script, but page 2 is missing the script – everyone just bounced when they hit page 2, even though they’re technically still on your own domain. I’ve seen site with a framed in element on the page cause some crazy numbers. If you click on an element in the framed in page and go to a page, either on your domain or on another domain you control – and the scripts don’t follow the click – that’s a bounce.
Cross-domain tracking is a big factor in bounce rates. If it’s not configured correctly, you’re going to see someone land on one domain, click a link to another – and get recorded as a bounce.
Here’s a practical example:
If you don’t have cross domain tracking set up in all three jumps above – jump one to jump 2 is a bounce, activity on URL 2 is not tracked, we will never know if the user spent money with us after they went to the second domain.
That’s important information to have, and you just inflated your bounce rate with information that is not accurate.
My Neighbors Are My Top Visitors?
I’ve written about using filters to get honest data here before, but we still see a lot of geographic activity come from a client’s own neighborhood. If you’re a mom & pop pizza place, that’s great; if you’re a resort in Orlando, and locals aren’t renting from you – so why so much activity?
It’s more likely you’re recording activity from your own employees, whether they’re in the office, at home, visiting a friend or sitting at the Starbucks down the street, if their IP address isn’t filtered, their traffic is affecting your data – and not in a good way.
I would say, as a company that does a pretty good job with analytics, troubleshooting takes up about 75-80% of the time we spend on analytics for our clients, the installations are easy, it’s answering the “why isn’t this working?” question that takes a ton of time.
Hopefully, my hints and tricks above have helped you cut out some of the investigation time and tweaking and you’re back to making money and serving customers in a jiffy.
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.