The Google Website Optimizer is quite possibly the best tool in the AdWords tool palette, simple because it will help you increase conversions easily, and increased conversions will do more for your internet business than any other metric. While it is a terrific tool, many PPC advertisers are not using it because they have not one but many landing pages, and they may have concluded that to test those landing pages they’d have to create a separate Website Optimizer test for each landing page. Not so. There’s a super-easy work around for this problem that may lead to doubling your profits.
For those of you who are new to the Website Optimizer, check out Google’s superb documentation and tutorials for the specifics of setting up a landing page test, or “experiment” which is what Google refers to landing page tests as. For the sake of this article, I’m going to use the example of a pay per click marketer sending ppc traffic to a dozen different landing pages that are standard, old-style squeeze pages with a bit of content and an offer of a great report in exchange for the visitor filling out a simple form. Each squeeze page has slightly different ad copy so that it matches the search phrase. In other words, the advertiser is sending all the clicks from the keyword “reverse mortgage” to a page with “reverse mortgage” ad copy, and all the clicks from the keyword “prime mortgage” to a page with “prime mortgage” ad copy. The words on these two example pages are different, but the advertiser is using the same header image, house image, form and other page elements.
This advertiser wants to test the combinations of these different elements, so they are doing a multivariate test instead of an A/B split test. For testing newbies, an A/B split test is when you test two completely different pages against each other. A/B tests are good if you are early in the testing process and have a lot of different things to test at once and you want results as fast as possible. Multivariate tests are a bit more complicated, as you are testing combinations of different elements. For example, an A/B split test might have a page designed by you running against a completely different page designed by your partner. In a multivariate test, you might be testing three different header images and three different form designs to see which combination of header and form converts the most visitors. In our example – remember the 12 ppc landing pages? -the advertiser is running a multivariate experiment.
The advertiser wants to test the form, the header image and the house image on all 12 landing pages but they don’t want to have to set up 12 different Website Optimizer experiments. What to do?
Here’s the secret: put all the landing pages you want to test in the same directory on your website, and then put the Website Optimizer code on all the pages in the exact same way. Let me be a bit more specific for the people who may never have actually set up an experiment. For a regular multivariate Website Optimizer experiment you put one type of code at the beginning and end of the whole page, and you put tags before and after the html that creates the different elements you want to test. So our advertiser is going to set up the optimizer code on all twelve pages the same way, and that is not going to have any effect on the different copy on each page because the before and after tags aren’t being put in anywhere near the ad copy.
The advertiser then loads all these pages on their website, confirms the optimizer code is in place the same way they would for a normal experiment, and launches the experiment. I have been running ongoing multivariate tests on one of my sites using this technique for over a year now. Just keep the optimizer code exactly the same on all the pages and it works exactly like a one page experiment.