The following post is a guest post. SEO Scoop endeavors to showcase and highlight guest bloggers, in the hopes of giving up and coming bloggers and SEOs a chance to shine. Please take some time to appreciate their efforts by visiting their sites, or by networking with them.

SEO Scoop does not necessarily agree with or condone any tactics mentioned by guest posters.


by: Lenny de Rooy

The Google Website Optimizer is a relatively new tool. You can use it to test different versions of your webpages, to see which one will generate more conversions. Bill Slawski discussed the patents of this tool and you can find even watch one of the introduction videos.

The tool is becoming quite popular, especially because it is free. However, I have noticed that there are still a lot of misconceptions when it comes to using the Website Optimizer. In this post I will describe the 5 most common ones.

Misconception 1: You'll need an AdWords campaign to use the Website Optimizer

Google chose to make the Website Optimizer part of the AdWords interface. Therefore, many people think you need to set up an AdWords campaign to use the Website Optimizer. However, it only means that you have to create an AdWords account. You do not actually have to run a campaign.

Google website optimizer

When you set up your AdWords account, just enter some fake keywords and ad texts. When you are being asked to enter your payment details, leave it blank. Your campaign will not start unless you have supplied your payment details, so now you can set up a Website Optimizer test without having to worry about AdWords.

All visitors who access your test pages will be included in your test. This can include traffic from paid search if you wish, but also from people who visit your site through organic search, links, and direct access.

Misconception 2: You'll only need to create a Website Optimizer account to use the tool

As stated before, you don't need an AdWords campaign. But you DO need a Google Analytics account to use the Website Optimizer. This is the only way the Website Optimizer can measure the results of your test.

It is not necessary to implement the Analytics code on all your web pages though adding it to your conversion page and the pages you want to test will suffice.

I wonder how long it will take Google to realize that it makes more sense to make the Website Optimizer part of an Analytics account, in stead of an AdWords account

Misconception 3: There is a big difference between an A/B-test and a Multivariate test

Let's start by explaining the two types of tests.

With an A/B-test, you create different web pages. The Website Optimizer will alternatively send visitors to page A and page B.

With a Multivariate test, you only create one webpage. Then, you tell the Website Optimizer which sections of this page you want to vary. For example, you want some visitors to see the page containing picture A combined with text B, and others to see the page containing picture B combined with text C. The Website Optimizer will rotate the content of the sections you indicated, so different visitors see different versions of the page.

ab testBig difference, right?

Actually, an A/B-test can do the same as a Multivariate test, and vice versa.

Don't focus on the way the test pages are put together. Just think about the eventual output.

In stead of a Multivariate test with the variants being pictures A and B, and texts A and B, you could run an A/B-test with four separate web pages: one with combination A-A, one with A-B, one with B-A, and one with B-B.

Alternatively, if the alternating sections you specify on the multivariate page consist of a large amount of html code, you can create two (or more) quite different pages, which will vary more than only in small elements. For example, you do not tell the Website Optimizer to vary just the code to display an image, but you tell it to vary the code of the full header div and the navigation div.

Of course, using the multivariate test in stead of an A/B test is not always possible, as this will depend on the way your webpage is coded. But as far as I can tell, there is no situation in which an A/B-test could not take the place of a Multivariate test.

Misconception 4: With an A/B-test, you can test only two versions of a page

Misconception 3 might very well be inspired by misconception 4. The name 'A/B-test' is somewhat misleading, as it may suggest that you can only test two versions of a webpage.

However, the test could have been called an 'A/B/C//N-test', as you can in fact test an unlimited number (well, almost) of pages at the same time. Just specify more alternative pages in your test set-up.

Misconception 5: With an A/B-test, you can test very different webpage-designs simultaneously

Actually, this is true. You can create two versions of a website that look nothing alike, and then test which one performs better.

This is not recommended, though. Say, one page clearly performs better than the other one. Do you know what caused that? Do you know what to tweak next on that page to further improve performance?

Only with step-by-step changes you will gain a better understanding of which page elements are responsible for the increase in performance. So although testing different pages is possible, make sure not to overdo it.


Author Bio:

Lenny de Rooy is a Website Usability Consultant at Tribal Internet Marketing. After work, she tries to find some spare time to apply her knowledge to her personal Alice in Wonderland website.

You May Also Like

11 Responses to “5 Common misconceptions about the Google Website Optimizer”

  1. MAcsSNAcs says:

    I tried setting up the AdWords acct, without the billing info, and when I click New Features to get the Website Optimizer link on the tab bar, click that, it just takes me right back to the "set up your billing info" page… Is there something I missed?

  2. SearchCap: The Day In Search, March 5, 2008…

    Below is what happened in search today, as reported on Search Engine Land and from other places across the web…….

  3. Gab Goldenberg says:

    Lenny, thanks for the link on the difference between ab and multivariate testing.

    Couple of things I'd like to share.

    1) The pros at SMX suggested doing AB for important issues first, then running multivariate. Not sure I understood why, but I guess it's so that you're at least 100% certain about the big things right off the bat.
    2) Good idea stating that a/b can go to n variations. That should be repeated as often as possible!
    3) The last three points are general stuff about testing, rather than Website Optimizer specifically. I'd love to see more from you, especially with case studies :) .

  4. Gab Goldenberg says:

    BTW, did you run this search: Are you familiar ab testing and multivariate testing?
    I wrote something a while back about monitoring logs to see who searches for something just before you get a new link, and this seems to be an interesting application.

  5. After setting up a fake campaign and choosing your login name & password, you should login again to Google AdWords. You are then being taken to the Campaign Management tab. In the green bar, there is a link to the Website Optimizer. There shouldn't be any questions about providing billing information.

    Perhaps you can send me a screenshot?

  6. Hi Gab,

    Multivariate testing is indeed most suitable for testing minor changes on your pages. You can use it to fine-tune the pages once you have figured out a good basic lay-out.

    You are right that the last three points are more about A/B and Multivariate testing in general. However, by making it a free service, Google has now made these kind of tests available to the general public, who have never before heard about them or used them before. So Website Optimizer users are the most likely ones to get confused.

    I am currently working on some interesting cases and will definitely post about them when the results are available!

  7. Hi Lenny,

    Great article, it's good to see the profile of multivariate and A/B testing being raised somewhat compared to where we were at this time last year!

    Just to let your readers know that commercial technologies for multivariate testing with a richer feature set also exist. For example, the Maxymiser technology can test variables on multiple pages simultaneously and continually optimize tests – showing proven best performing versions to future site visitors.

    We also provide account management and managed services around what to test in order to generate the best sales uplift. From our experience, a good part of the challenge clients face is knowing what to test; the technology is only part of the puzzle.

    Kind regards,
    Alasdair

  8. [...] Lenny de Rooy, wrote a guest post at SEO Scoop about 5 misconceptions of Google Web Optimizer. It goes slightly beyond just GWO itself and into testing [...]

  9. Today, Google announced that the Website Optimizer is out of Beta (http://adwords.blogspot.com/2008/04/website-optimizer-now-out-of-beta.html)

    Most important update is that you no longer have to open an account through your AdWords account. The Website Optimizer is now one of the services offered that can be used with a Google account only.
    Sign up via http://www.google.com/websiteoptimizer

  10. Tonee says:

    The most important factor that must never be forgotten when dealing with Google AdWords, is that taking shortcuts doesn't pay off! Relevance is so important — just give people what they are looking for! After all, this is the primary differentiating factor between Google AdWords and other more traditional forms of advertising!

  11. Eric says:

    Thanks a lot!

    me and my partner were a bit worried this optimizer was for only Adwords. Glad I found this blog post.

    -Eric
    http://ericsender.com
    http://mydrupaladventures.com