The number of free programs Google offers to help you improve your website and generate valuable data is truly astounding, but one of the most underutilized programs in this suite is the Google Website Optimizer. If you aren’t currently using this great free resource to improve your website, read on for more details on how to incorporate this tool into your web development arsenal.
What is Google Website Optimizer?
Basically, Google Website Optimizer is a free tool you can use to carry out split tests on your website. We’ve talked about split testing a lot here on the Single Grain blog, but to refresh your memory, split testing is an incredibly valuable process that allows you to compare specific elements on your site’s pages in order to make your overall website more effective.
For example, suppose you run a long form sales letter website that sells a $7 report with a more expensive product as an upsell on the backend. You’ve built a beautiful site and you’ve got a great product, but you aren’t sure if your prospects will be more motivated to buy if you advertise using “Headline A” over “Headline B.”
Sure, you can make your best guess and hope that your readers respond the way you think they will – or, you can actually determine which pitch is more persuasive by testing one against the other!
Using a split testing script like Google Website Optimizer, you create two versions of your sales letter – each featuring a different headline that you want to test. After uploading your new pages and validating your script, Google Website Optimizer will launch both versions of page in random order as people visit your site, allowing you to determine conclusively which headline variation is more persuasive to your readers.
Should You Use A/B Split Testing or Multivariate Testing?
The scenario described above – in which two headlines are tested against each other – is what’s known as an “A/B Split Test”. That is, variable A and variable B are being compared, while all other elements on the site remain constant.
A/B split testing is a great place for most webmasters to get started, as it’s easier to observe conclusive results when only a single variable is being changed on your site. Unfortunately, though, generating results with A/B split testing can be extremely slow. Because there are an unlimited number of variables on your site that you could test, only performing one split test at a time means that it could take a while for you to hit on the variables that lead to the biggest improvements in your website’s efficiency.
Multivariate testing is the more advanced version of this process, which allows you to test multiple variables at the same time. It certainly allows you to uncover results faster than A/B split testing, but the complexity of these tests means that they aren’t well suited for beginners. When testing multiple variables against each other, it can be difficult to determine which specific variables led to the improvements you see, which is why it takes a more experienced tester to make the right conclusions based on the data that’s generated.
The bottom line is this… A/B split testing is slower to achieve results, but easier for beginners to implement than multivariate testing. If you haven’t done much split testing before, start with Google Website Optimizer’s A/B split testing feature until you’re confident enough in your skills to move on to more complex testing.
Setting up a Google Website Optimizer Test
To get started with Google Website Optimizer split testing, navigate to http://www.google.com/websiteoptimizer and log in with the same Google account you use to access Google Analytics data for your website. Upon logging in, you’ll see the details of any current tests you’re running, as well as a link labeled, “Create another experiment.”
Clicking on this link brings you to the first page in the process, where you’ll select between running an A/B split test or a multivariate test:
As you can see, Google lays out many of the same recommendations for choosing either type of testing as described above. For the purposes of this article, let’s assume you’re a beginning webmaster who wants to start with an A/B test.
The next screen that appears asks you to take a few preparation steps before proceeding with the test. They include:
- Identifying the specific page you want to test (for example, your home page, your checkout page, your upsell page, etc)
- Creating each variation page you’ll use in the test and uploading them to your site (following the example above, you’d create two versions of your home page, each displaying a different headline version you want to test)
- Determining your conversion page (aka – the page a visitor lands on that confirms that a conversion has taken place)
Once you have this information ready, check the box labeled, “I've completed the steps above and I'm ready to start setting up my experiment” and click “Create.”
On the next screen, you’ll enter in data related to your test, as determined by the preparatory steps you took earlier in the process, including a name for your test, the specific URLs of your variation pages and the URL of your conversion page:
Finally, on the last page, you’ll receive snippets of code to add to each of your variation pages and your conversion page, along with specific instructions on where to place this code within your page’s HTML. After uploading all of this information, press the “Validate” button in Step #4 and Google Website Optimizer will confirm that your test has been set up correctly.
If everything checks out, your test will begin running right away. Check back periodically to view your progress, but remember that it’s a good idea to collect data on at least 100 completed conversions before drawing any conclusions about your data.
How to Get the Most Out of Google Website Optimizer
Now that you know how to set up a successful split test within Google Website Optimizer, how should you get proceed in order to get the most out of this useful program?
According to Cameron Chapman, writing for web analytics giant KISSMetrics, the best place to start is with your sales funnel:
“The best way to figure out what to start testing first is to look at your current conversion funnel. If 70% of people aren’t making it past your homepage, start there. If you’re losing half your visitors at the shopping cart, then test and optimize that first.”
Once you’ve made your sales funnel as effective as possible, consider experimenting with site-wide changes or specific elements in the sales process (for example, your headline, sub-headline, “Buy Now” button and more). It’s also a good idea to take a look at what your competitors are doing. If they’re testing new site designs and structures, there’s a chance they’ll work for you as well.
However, the most important advice is to simply start testing and keep testing continuously. There’s no limit to the number of elements on your website that can be tested, so there’s no reason you shouldn’t have a test running at all times. As your split testing skills improve, try to transition to multivariate testing as well in order to make your website more effective, as quickly as possible.
Image: Tamar Weinberg