Google Analytics and Umbraco: a not-so-perfect pairing
Just because everything is so cozy on the surface, doesn’t mean that everything is perfect. Using Google Analytics to A/B test your Umbraco site often presents a few challenges that will bog down the test on your end and create UX issues for visitors on the other side. If you run a test using Analytics, you’ll notice:
- You’ll need some programming knowledge- Umbraco is great for making content changes, but for anything other than that you’ll need a developer (refer to section 1.1 for which tasks will require a developer’s help).
- The redirect issue- one major hurdle is the fact that GA Experiments relies on redirects to send visitors to the different variants that you’ve set up. In Umbraco that often requires duplicating an entire page. When you’ve got a page that has many child pages, you now have to manage the branches of a rapidly growing tree of pages. Redirects is a UX killer, and more requests to the server could result in slower loading times.
- No feedback while experiments are running- experiments could up take up to two weeks or more to return statistical significance. This is especially annoying when you’ve obtained confidence in one variation already, but 50% of your traffic is still sent away. In rare occasions, it could cause you to lose out on sales.
- Context switching (between Umbraco and GA)- when managing multiple variations, you’ve got to constantly be switching back and forth between Umbraco and GA to paste URLs, copy JS code, and paste it back into Umbraco. All of this extra labour works up a good sweat and wastes a lot of time. If you choose to automate this process, you will save a lot of time.
- Plus Umbraco’s A/B testing headaches- on top of GA’s issues, Umbraco’s A/B testing drawbacks are still present. No concept of a variation, variations showing up in menus, and possible duplicate content issues further complicate only add to the difficulties presented with A/B testing when you use GA with Umbraco.
So aside from needing someone with programming knowledge, you need someone with access to your GA account and the knowledge of GA Experiments to actually set up and run the test. Duplicating pages leads to more work on your side and could harm your UX by slowing down load times, breaking your menus, or irking users that land on your page only to be whisked away to a variation immediately upon arrival. In a world where nearly half of consumers expect a site to load in 2 seconds or less and will abandon a site after 3 seconds, speed is killer (source).
If integration with Google Analytics is important to you, but you would like to administer your experiments more effectively, you may use plugins such as uSplit that remove most of the technical obstacles to A/B testing. Automating most of the laborious processes will streamline the process and allow you to run experiments with as little overhead as possible.
Google Analytics is a free, relatively simple, and powerful tool for gaining insight into your visitors’ behaviour. Using the Experiments tool to run A/B tests gives you the ability to find out what improves important metrics the most. Despite how great the tool is, running tests with it on your Umbraco website means you’ll encounter a few technical obstacles that will make your life a bit harder and might lead to a slightly downgraded UX. If you’d like to eliminate those headaches with a simple-to-use tool, then check out uSplit, a free plugin that does exactly that.
For more information on how to run killer A/B tests with visual tools, check out our article on Optimizely and Visual Web Optimizer.