Category: UX advice / Research articles
I’ve been using Google Analytics (GA) as a tool in my evidence-based UX design approach for several years—particularly when helping clients redesign their websites. It’s a great source of information as all the sites I’ve worked with already had it installed and gathering data.
However for all the masses of data you can find in GA to help with digital marketing, there actually isn’t a lot that’s truly useful when it comes to helping you make decisions around UX design. Luckily you can save yourself the hassle of staring at screens full of numbers and trying to understand what it all means, and take a look at the actionable data I use.
(Exactly what is most interesting can vary from site to site but the following are the most common starting points I use for finding patterns.)
It’s important to look at the same time period across what you record from GA, so the data all matches. It should be of a length that won’t be warped by any days or weeks with odd spikes or dips in traffic. Ideally I’d look at the most recent quarter or three months, which will average out any noise.
If there has been a major change of design in the website in the last quarter so that older data refers to something that no longer exists, then studying the most recent month should be enough.
If the website has any goals or conversion funnels set up (both found under Conversions > Goals) then this should be your first port of call to check. Funnels are most useful if the site is built around a fixed set of steps that users must go through as they clearly show if people are making it too that next step or not.
You’ll find out whether the website is converting people into buyers or leads at a steady rate (or at all). If it is changing over time is it in a positive or negative direction? If things are improving perhaps a light redesign is all that is required but if they’re getting worse then something more radical could be considered.
A funnel will show you the rate at which steps are converting and importantly which steps are converting the least well. These will be ones that need the most effort and improvement.
GA contains some information about who your audience are, and in particular you can see what age brackets your users fall into (Audience > Demographics > Age). If you have a goal set up you can see how your users convert by age in the right hand columns and learn if this differs for a particular age group.
It’s always handy to know if the real-life users match up with who you think the site appeals to and if that target audience are actually likely to buy. I’ve also written about using GA to create basic personas or user statements to summarise who you’re targeting and help with recruiting people for user testing.
There’s quite a lot of information available in GA about which device your users have when browsing your website. At the very least you should find out what your overall split of mobile, desktop, and tablet users are (Audience > Mobile > Overview). Again if you have a goal in the right columns you can see if users are converting at different rates on different devices: it’s not unusual for them to vary.
At the next level you can find out the popularity of operating systems within those devices such as Windows v Mac or Android v iOS (Audience > Technology > Browser & OS). Also in this section you can see which browsers people are using on those operating systems—as most browsers are available on different devices. Go to the Operating System tab and select the secondary dimension of Browser to get them split out so it’s clearer to understand.
All of this will make it easier to identify what browsers your designs need to support and any that you should be optimising for if they feature in large numbers. It also helps for making your user testing accurate and means you’ll know what to test on when the designs are built.
If a device has a lot of users but isn’t converting very well then this is somewhere to focus more of your research, and suggests that the current design needs lots of improvement.
I study how popular the different sections are that appear in the website’s navigation (if there are many then I take the top level of nav). This can be found by going to Behaviour > Site Content > All Pages and looking at the unique pageviews. I record the percentage score for the relevant URLs that is shown next to the absolute number.
This screen is also useful for finding the average time on page and exit rate columns. These numbers don’t tend to mean much in isolation but if any of these are much higher or lower than the average then I will make a note of this.
This helps with nav design by giving an indicative idea of which parts of the site are most popular—if some are low it makes me question if there is sufficient interest for them to appear on the main navigation. In addition I might spot sections that have higher traffic but currently aren’t in the main navigation when they could be.
Very high or low average times on page and high exit rates are things to look into further with qualitative research to find out why this is and whether users are uninterested in the content or getting stuck.
A look at the locations of users (Audience > Geo > Location) will tell us which country they come from. Typically this is fairly well understood by clients but it can help us see if any countries are more popular than we had expected and whether we need to design to support that audience too.
Finding out the split of new versus returning users (Audience > Behaviour > New vs Returning) can help you understand who you are speaking to. It’s common for there to be more new users than returning but if your number of returning users is particularly high (over 30% as a rough guide) then there may be time-saving functionality like ‘recently viewed’ or account log in that needs promoting on the landing pages.
Finally, if you have specific events set up triggered when users do an action such as click (Behaviour > Events) then these will typically tell you the popularity of in-page elements and buttons, and which ones users are most drawn to. You can make sure your new designs maintain them and rework the ones that don’t get usage.
The above areas are ones that I typically include in all of my analytics reports for clients, along with findings from heatmaps and watching visitor recordings. In fact I’ve put this structure together (including some handy tables) in a Keynote and Google Slides template here which you can use to save yourself some time in reporting.
Sign up here to get a guide to my favourite (mostly free) tools for evidence-based designing. Plus a massive, advice-filled reading list.You'll also get UX articles & content emailed to you once a week. Your email is never shared. Unsubscribe at any time.