I’ve been talking to a new client this week, who’s launching a new service. Well, an existing service being delivered in a new way. Part of my investigation into how we’re going to help them promote that service online was to look at how it had been performing historically.
So I’ve been poking around in Google Analytics.
They’ve had great visitor numbers over the years, a nice split between new and returning users, and engagement metrics are fantastic… like, really fantastic… a bit too fantastic…
… bounce rate is mad-low, pageviews are crazy-high, and pages/session are just kookoo bananas – I think we have a problem.
Google Analytics 101
A while ago, I wrote “Google Analytics – The Basics”. It was a post about some of the basic data Google Analytics has to offer, tailored to GA noobs. I went over the Audience, Acquisition, and Behaviour tabs, but here’s a little bit more detail on some of the metrics:
The number of visitors to the site over the set time period. If a user returns to the site multiple times, they will be counted multiple times, but you can see “New Users” for the number of first time users within the set time period.
Once known as visits, sessions are defined by Google as “a group of user interactions with your website that take place within a given time frame”. One session is what a user does when they come to your site:
Enter via the homepage >> read a blog post >> complete the contact form >> exit
That’s a session.
The total number of pages viewed over a given time period. In the example above, that session would register three pageviews (or four if the contact form has a “thank you” page). Multiple views of the same page will count as separate pageviews.
This is the average number of pages viewed inside a session across all users in a given time period. So, if one user visits the homepage and then reads a blog post, that’s two pageviews. Averaged with our four-pageviews example above, pages/session would be three.
Avg. Session Duration
The average length of a session on your site. So, if user one is on the site for one minute, and user two is on for three minutes, the average of those sessions would be two minutes.
Calculated as a percentage, this shows the sessions in which a user has left your site after visiting only one page, with no interaction.
If these are some of the most well-used metrics in Google Analytics, let’s have a look at where errors can occur, and what we can do to troubleshoot them.
Google Analytics Woe 1
You’ve logged into analytics to see what your sessions look like, but you’re presented with a flat line.
What to do: Check you’ve got the code installed, and check it’s in the right place. Go to your Admin panel in GA…
… then select Tracking info from the property list, and then Tracking code:
“Copy and paste this code as the first item into the <HEAD> of every webpage you want to track.” << THIS IS VERY IMPORTANT.
Google Analytics Woe 2
You’ve logged into analytics to see what your sessions look like, but you’re presented with a low line.
What to do: You need to check you’ve got the tracking code installed on every page of your site. You should be able to see a list of your web pages under Behaviour, Site Content, All Pages:
Google Analytics Woe 3
Bounce rate is mad-low
Average bounce rate will differ from site to site depending on its purpose, it’s performance, and its popularity, but if you’re in single digits, you’re definitely going to need to check something’s not up.
What to do: Check the code’s not been installed twice. This can happen if a couple of people have been working on the site, or (as was the case with my client) if the code is found in Tag Manager as well as in the site’s <head>.
I use a handy little tool from Google; Tag Assistant, which helps to troubleshoot the installation of Google tags. It’ll show you if there’s two (or more!) instances of your tracking code.
When I removed my client’s duplicated tracking code, their bounce rate went a little something like this:
… and their pages/session went a little something like this:
Google Analytics Woe 4
If site sessions dramatically increase, and you’re not running a killer campaign, you need to check your Referral traffic. If you’re seeing sources like semalt.com, dadorar.com, or buttons-for-websites.com, with zero average session duration and 100% bounce rate, you’re suffering from referral spam.
What to do: Moving forward, you can use a filtered view in GA. Before you do so, it’s a good idea to create a new view specifically for your filter, so you have access to raw data.
To retrospectively remove spam referrers, you need to create a custom segment.
Click on + NEW SEGMENT, name it, then under Advanced, select Conditions, then choose exclude, source, matches regex, and then list the spam sites enter those pesky sites using a pipe (|) to separate each one.
Read Browser Media’s guide to removing referral spam in Google Analytics
You can’t analyse user behaviour without user data
This really is just the tip of the iceberg when it comes to fixing errors in Google Analytics, but getting your head round these four is a good start. I mean, it’ll ensure you’re collecting accurate user data! With a reliable snapshot of how people are using your website, you can begin to understand where the pain points might be, and that’s when you can start making changes to improve user experience.