How to measure UX test improvements using Google Analytics
See how we do it, using the WhatUsersDo blog as a guinea pig.
A recurring issue (even among our customers) is that most of us simply suck at measuring the business improvements achieved through UX testing.
When I ask people (customers and non-customers alike) whether a round of UX testing yielded positive results, I get vagaries like:
- 👍 Yeah! Everyone was really impressed by the issues we found!
- 😃 Testing was so quick and the quality of the feedback was surprisingly good.
- 🤔 What do you mean? Like… did traffic increase?
Being a hard-nosed, business-minded marketer, I’m never satisfied with such answers—and you shouldn’t be either. Cultural change is good. Insightful user feedback is good. Traffic is good.
But all those things are the means—they are not the end. Executives care about the bottom line.
When I say “results”, I mean money or a direct proxy for money (e.g. software signups). I know the pure UX designers among you might turn your noses up at that.
- 😻 It’s like… about the experience, dude. We do it for the love, not the money, man.
That’s all well and good (and true). But I’ve seen first-hand that if you don’t show commercial returns to the business, it’s only a matter of time before investment in research and testing gets cut.
I’ve decided to share my process for running UX tests on the WhatUsersDo blog and getting tangible business results out of the insights gleaned. Tangible business results like:
266% increase in the number of monthly free trial signups via the blog
300% increase in the number of monthly content downloads via the blog
885% increase in all monthly goal completions via the blog
Once changes based on UX testing had been made, the numbers above fluctuated month-to-month—sometimes going a little higher and others a little lower.
But the overall trends remain the same, with free trial conversions moving into double figures (from lower, single digits) and all goal completions increasing dramatically.
I chose to benchmark against August because up until that point, my focus had been on creating high-quality content, which attracts relevant people (and keeps them coming back).
We’re a niche within a niche (remote UX testing, within user experience as a topic). Building a reputation and *relevant* audience (without paid acquisition) is no joke.
Again, without resorting to bad UX or spammy tactics, here’s the growth in traffic we managed between March 2016 (3 months after I’d joined WhatUsersDo) and September 2016:
Once I’d created enough momentum traffic-wise, I began focussing more on making the most of that traffic, through the power of great UX.
Using my process as a template, I’m going to share:
How to set tangible, measurable goals for UX testing
How to write UX testing tasks that will reveal obstacles to achieving your goals
How to measure before-and-after performance—so you can quantify and attribute improvements[/color-box]
All improvements shown here are as a result of changes based on UX testing alone. We spend exactly £0 on promotion of our content. We serve 0 pop-ups on the blog and don’t try to spam people into doing anything.
Setting tangible, measurable goals for UX testing
Simply “improving the user experience” is too vague. Every experience comprises multiple interactions and the key is to pick a specific interaction you’d like to improve.
Don’t overthink this part. For example, the interaction you choose could be as simple as increasing click-throughs to a page with a high conversion rate.
Just think of the closest link between the thing you’re testing and revenue, then aim to improve the performance of that link by some percentage points.
For example, it’s damn difficult to get you to go directly from reading this blog to paying for the WhatUsersDo platform. A handful of people have done it in my time here, but it’s very unlikely.
However, people do go from reading the blog to:
- ✅ Signing up for a free trial
- ✅ Downloading an asset
- ✅ Signing up for the newsletter
And some of the people who do those things become paying customers. So, these were the links I identified between the blog and the business—conversions.
Here are some parameters to help you select effective goals for UX testing:
📈 Your goal should ideally be quantifiable in numbers
🕗 Your goal should preferably be a metric you’re already measuring (so you can benchmark)
💰 Your goal should be no more than 2 steps removed from revenue—i.e. Goal -> Next Step -> Revenue—and, ideally, only one step removed[/color-box]
Now we know the UX metrics I wanted the insights from UX testing to help me improve—free trial signups, asset downloads and newsletter signups.
How did I design my UX tests to achieve those goals?
Writing UX testing tasks that will reveal obstacles to achieving your goals
So, if you haven’t read our 8 tips on writing incredible UX testing tasks, go do that right now. It will give you a good foundation for this discussion.
Above is a screenshot of the tasks I set users in the first UX test I ran. My approach was simple:
I identified all the starting points from which people might arrive at the blog (homepage, search, social etc.), then picked one for my test script
I identified the goals people might want to complete (and the goals our business needs people to complete) via the blog
I set scenarios which involved users finding their way from the relevant starting point to the end point of a goal completion (without explaining to them how to do this)
As you can see from the screenshot of the UX testing tasks, users needed to:
- ➡ Go from the homepage, to the blog, to a free trial signup
- ➡ Go from the homepage, to the blog, to the newsletter signup
- ➡ Go from the homepage, to the blog, to an educational content asset (downloadable or not)
My hypotheses were that of the users who enjoyed the content they found on our blog:
Some would want to try out the platform (free trial)
Some would want more content of a similar ilk and quality (newsletter)
Some would be curious about UX testing, but not confident about its finer details (educational content)
Upon running two rounds of UX testing, these hypotheses were proven accurate (along with some others that I hadn’t thought of).
The results of testing also revealed issues users falling into group A, B or C might face, while browsing the blog.
User wants to learn more before clicking free trial
For example, based on this user’s behaviour, I changed the menu items in the top nav.
The gentleman twice hovers over the option to view the free trial page but chooses not to do so—saying at one point, “I’m not gonna do that.”
So, I quelled my marketing thirst for leads and decided to replace “Success Stories” (salesy) with “What Is UX Testing” (educational), in the blog’s top nav.
The result? A significant reduction in the number of people clicking the free trial link—but a significant increase in the number of people signing up for free trials.
I know. It’s counterintuitive. I still don’t fully understand why these two changes occurred together but I have my hypotheses. Perhaps more people felt informed enough to decide (definitively) whether or not UX testing is for them.
Another user validated my hypothesis that the sidebar on the blog was being wasted. It was occupied by stock widgets and redundant information, that neither educated users nor supported our business.
User goes to sidebar to find educational content
This user was not the only one to behave this way, so I knew I had to find a better way of solving her problem.
I added a combination of educational content with decent images, and downloadable assets to the sidebar. I also removed or compressed extraneous information.
As you can see, the tasks I set drove users down certain paths which made it necessary for them to reveal obstacles in the way of their goals (and mine).
For example, if your goal is to find out whether users can successfully complete “section X”, set a scenario you know will involve them passing through “section X”. Just don’t tell them what you know and don’t tell them what to do when they do arrive at “section X”.
Measuring the impact of changes based on UX testing (using Google Analytics)
In terms of analytics, there is no better expert I know of than Jill Quick—the grand sage of Google Analytics.
Before you read my super-condensed version of how to measure goals, read as much as you can on Jill’s website. Then pay for her consulting services. Honestly, you’ll thank me.
In terms of the UX tests I ran on our blog, there are 3 main ways I measured performance:
Google Analytics goals
Google Analytics segments
Google Analytics custom reports
Google Analytics goals are parameters you can define to let Google know how to measure commercially important activities on your site (e.g. a sale or free trial signup).
Google Analytics will then be able to tell you how many times that activity has occurred, who completed the goals, where they came from etc.
You can find the area for setting up goals in Google Analytics by going to: Admin -> View -> Goals.
This is good for an overall picture. But I can’t see, off the bat, only people who completed said goals via the blog.
- The goals reporting tells me, “This many people did X, coming from Y, using Z device…” etc. But it won’t automatically categorise people who did “X” for me.
- What I really want to know is, “How many people did X, via the blog?”
To answer that, I need segments—think of segments as stencils you lay over your raw data, to reveal only people who fit within certain conditions.
The shape of my “stencil” was designed to show only people who completed goals via the blog, without also visiting pages for our case studies. This was to prevent duplicate reporting… remember, we’re focussing on the blog here but I also create other types of content for WhatUsersDo.
Google has written a short guide on how to create segments. Here’s what mine looks like (with arrows indicating customisation options and areas of interest):
We’ve covered segments, which help with slicing up my data… but how do I see all the information relating (specifically) to my blog UX testing project, in one place.
To do that, I need custom reports.
Think of a custom report as a filing cabinet that lets you group together data you care about, regarding a specific project. You can customise the item about which you’re seeing data, as well as the kinds of data you see about that item.
For example, you can choose to see only the “page views”, “time on page” and “bounce rates” for the “Resources” pages on your site, in one custom report.
Here’s what my custom report, which shows the performance of our blog, looks like:
And here’s how I’ve set up the customisation in the back end (although, I’ve also added several filters which you can’t see in this image):
And there you have it! Now you know how I:
Set goals for running UX tests on the WhatUsersDo blog
Design tests that help us improve our performance in achieving those goals
Accurately measure the impact of changes made as a result of UX testing
Please note, this article was originally published on the WhatUsersDo blog. We’ve re-edited and updated the content.
Timi is a passionate creative and meticulous business strategist. He currently designs and executes the content strategy for PatSnap’s marketing programme. Timi is the former senior writer and content strategist at WhatUsersDo.