As a copywriter I’m often surprised at which words actually sell. The wrong words can reduce the sales of a product. The right words can increase sales. Awards aside, the truism of advertising given voice by one agency is “if it doesn’t sell, it isn’t creative.” Nowhere is this truer than in landing pages.
If your goal is increased conversions, let the marketplace decide what works. How? By testing. Testing provides actionable data that translates into increased ROI. Testing takes the guesswork out of landing page copy and design, replacing it with results.
Landing page testing and optimization is now within everyone’s reach. Google has made their Website Optimizer freely available through the AdWords interface.
Split Testing and Multivariate Testing
Online advertising commonly uses two types of testing: split testing (also known as A/B testing) and multivariate testing.
Think of split testing as a competition and “may the best man win.” In split or A/B testing, you are simply testing one element against another. Half the traffic goes to one version of a landing page, and half goes to the other. Whichever converts best gets the full-time job.
This is a tad simplified, as you could have 3 or 4 landing pages vying for domination and it would still be considered a split test. But you get the picture. What, then, is multivariate testing?
Multivariate testing is the testing of more than one variable at the same time. Statistical analysis determines the optimal variables in part and in sum. This level of testing reveals which variable outperforms the others, as well as which combination of variables is optimal for the desired outcome. In the case of a landing page, this outcome would be conversions.
Google Website Optimizer allows testing of up to 8 landing page sections, 127 variations per section, and up to 1,000 total combinations. Theoretically, you could get similar results with a split test – if you were willing to test 1,000 different landing pages.
But multivariate testing is also more efficient than split testing in another critical aspect – the quantity of traffic needed to yield statistically valid results. As you can imagine, a staggering amount of traffic is needed to test a thousand combinations. With multivariate testing, the mathematical kung-fu that is statistical analysis will yield valid results while an equivalent split test is still chugging along and draining your budget.
The Testing Process
Test. The results could surprise you. Rather than running a split test on 1 or 2 variables, run a multivariate test on all critical page elements. Let the marketplace vote with its dollars and evolve a conversion-generating ROI machine.
And now for the tool. Take a tour of the available multivariate applications and you’ll find that cost probably takes all but Fortune 500 companies out of the game. Google Website Optimizer is the exception – it’s free – and in almost all cases is as powerful as (or more powerful than) the commercial stuff that will cost you the equivalent of your first-born child.
Always go into testing with clear conversion goals. Whether conversion means a purchase or a newsletter subscription isn’t important. That it’s a clearly defined goal is. Make sure conversion is the thrust of your landing page and that the results can be tallied.
The two key elements in testing are the test page and the conversion page. The test page is the landing page you want to test. The conversion page is the page the user reaches once they have heeded the call-to-action. This could be a download page or a “thank you” page. The key is that the conversion page can only be reached once a conversion has taking place. It defines the conversion.
Now take a hard look at your landing page copy and graphic elements. Decide what to test and brainstorm some variations.
Deciding What to Test
What should you test? In a perfect world, the answer would be “everything”. In the world that we inhabit the answer is “test the most critical elements.”
So what are the most critical elements? What, exactly, in a landing page makes the most difference in terms of a conversion? The answer to that is “it depends.” In general though there are some elements that can have a huge impact on conversion rates:
- Headlines and sub-heads
- Assurances and guarantees
In any type of copy headlines are critical. With those precious eight seconds that you have before the user bails, you have to convince them that they are in the right place and that they should continue. This is the job of a well-written headline or two. The headlines and sub-heads should also work together to summarize your offer – most people scan copy rather than reading it. Make sure that scanners get the message.
Your call-to-action is your “buy” or “subscribe” button. The wording is critical. Imperative verbs and a description of what happens next will increase the odds of success. Experiment with copy and graphical elements here. The POA (point of action) is also where you will want to place and test guarantees and assurances.
Images. Test which ones work and whether you need an image at all. Copy carries the day on landing pages and images are often only a distraction. The exception can be product shots, but you can only be sure by testing.
How many times have you abandoned a sign-up or checkout because the form was so intimidating? Lots, I’m guessing. Consider this and always test forms. Ask only for information that you absolutely need. Test variations of form design and copy as well as form placement on the page.
Lastly, test your copy. Picture your ideal prospect and write to them. Come at it from several angles and check the results. Should you use long or short copy? I don’t know. Test it.
Setting Up a Test in Google Website Optimizer
Once you’ve decided which elements you want to test, you’re ready to set up the test itself. If you’re comfortable cutting and pasting HTML markup, the process is quite simple.
The control script goes before the HTML closing head tag on the landing page. The tracker script goes before the closing body tag on both the landing page and the conversion page. The section script is placed both before and after each of the elements you are testing along with the variations of those elements.
That’s it. Validate the code through GWO and you’re off to the races. Make sure you test long enough to get valid results. Google has a calculator to help you figure out how long is long enough. Now check the reports and see what worked.
GWO reports are fairly intuitive. The report screen contains two tabs: the Combinations tab and the Page Sections tab.
The Combinations tab lays out the performance results for different combinations of tested elements. The combinations column on that page lists the results of your control page, followed by the best-performing combinations in order. You also get data that tells you the probability that a specific combination will out-convert your control page, along with an estimate of conversion rate for each combo.
The Page Sections tab gives you a relevancy rating for each element, i.e., which tested elements had the biggest impact on test results. The combinations page tells you which elements in concert produced the best results, but this section ranks the elements themselves.
Conclusion and Resources
Testing is vital to any campaign. The Google Website Optimizer is a free, multivariate testing tool that, used properly, can guide you towards increased conversion rates as well as a higher return on investment. You may be surprised at what works, what doesn’t, and how seemingly small details can have a disproportionately huge impact.
The Google Website Optimizer page (http://services.google.com/websiteoptimizer/) has a demo and a quick start guide, and is the resource of choice for all “how to” type information.
Future Now (http://www.grokdotcom.com/googlewebsiteoptimizer/) offers some great resources to help you get started.