August 10, 2017 | Sarah Danks
Split Testing for Conversion Rate Optimization
In the ever-so-competitive world of paid search, it’s not enough that you have custom landing pages. No. Just as you continually search for ways to improve the optimization of your website, so too should you be trying to constantly improve your PPC landing pages’ conversion rates.
Two ways to test landing page components for conversion rate optimization (CRO) are:
- A/B testing (also called split testing), and
- multivariate testing.
Depending on a few factors, you can determine which type of testing works best for your clients’ goals and budget, and your own capabilities.
That being said, you need to consider a few things before launching into a landing page test:
- How easy is it to make changes to a landing page? Do you have an on-site developer or are you going through the client’s webmaster for every little change? How time-consuming is the process of making updates?
- How much management time do you have per month on this PPC account? Will a lot of testing (design, development, tracking, etc.) eat up too much of your monthly budget?
- How long is the PPC campaign running? Will it span enough time to gather data, implement changes and then run long enough to be beneficial?
If you’re running a very small account that doesn’t get a lot of traffic per month, then A/B testing might not even be worth it. If, however, you have a client that’s spending enough per month — and receiving enough traffic — to warrant some testing for CRO purposes, then have at it!
But, what’s all this “testing” mean, anyway?
What’s A/B Testing?
Here we’ll talk about A/B testing, since we typically start off any type of landing page testing with this method. Quite simply, an A/B test is one in which you pit two versions of something against one another to determine a winner (based on conversions, since we focus on lead generation for our clients).
In our PPC tests, the same types of visitors use the same types of keywords, but the traffic is split between two versions of the same landing page. Once we get a winner from our split test, we can then move on to multivariate testing, to further hone the landing page’s performance.
Below is an example of an A/B test we did on a client’s PPC landing page. Here we focused on the contact form as our test subject.
As you can see, in our “A” test we had the form laid out on the right-hand side.
Landing Page “A”
And for the “B” test we decided to make the form far less prominent; visitors had to click on “Request Info” to bring up a lightbox with the form in it.
Landing Page “B”
In this case, it goes to show that having a prominent call-to-action/actionable item front-and-center on a PPC landing page pays off — the “A” landing page outperformed the “B” landing page by 106%!
Sometimes you might be testing something that *seems* intuitive, but you really never know until you’ve gotten your test results back.
Only Test One Thing at a Time
One thing to keep in mind while A/B testing: you must only test one item at a time. In the aforementioned example, while the design layout is different — to accommodate the form differences — the copy is the same. If you have separate copy, design, imagery, etc. you won’t know which of all the elements being tested are working best when you get the results from your split testing.
Test too many things that’re different between your “A” and “B” landing page versions and you won’t know which — or what combo — of the elements the visitors preferred.
“Winner” of A/B Test: Move it to Other Marketing
A/B testing your PPC landing pages to determine a winner can also help with other forms of marketing. Once you’ve discovered your highest-converting landing page from the split test, you can then migrate that landing page over to other forms of marketing (direct marketing, email marketing, etc.).
Will every form of traffic act the exact same way your PPC traffic does? Maybe not…
…but at least you know you’ve got a good, proven layout for conversion on at least one platform.
Also, by knowing which keywords, ad language and landing page copy convert the best, you can massage your website’s content to reflect the messaging that works the best.
Oftentimes, what works well for converting PPC traffic also works well with SEO traffic.
How Do You (Technically) Perform an A/B Test?
The technical aspects of A/B testing are fairly simple. In both Google AdWords and Bing Ads you can test manually by simply taking one set of ads and duplicating it. The duplicate set will of course point to Landing Page B (while the original goes to Landing Page A). Voilà.
We use Visual Website Optimizer (VWO) to run our A/B and multivariate tests. By adding one line of tracking code, we can easily pit two landing page versions against each other to determine the winner.
Plus, VWO’s code doesn’t slow the page/site load time, which is key for good UX!
There are many other services (some are free; some aren’t) to run your split tests. Here are a few other options for 3rd-party A/B testing:
- Convert Experiment
- Google Analytics Experiments
“Always Be Testing”
Split testing is a simple but very important step in search marketing. Of course, just because you found a “winner” in an A/B test doesn’t mean you’re done testing. Testing should be a continual part of any PPC management campaign — whether it’s for keywords, ad copy, colors, lead generation forms or any other landing page components.
Bryan Eisenberg and John Quarto-vonTivadar both said, “Always be testing.” I also like what David Ogilvy has to say about it:
“Never stop testing, and your advertising will never stop improving.”
So — go forth and test!