I can’t make my mind up how to start this post. Do you think
more people will read it if I start with:
How many hours have you spent in meetings debating which
graphic will make your web site make more money?
Maybe I should start with:
Google Website Optimiser takes the guesswork out of making
decisions about your website.
Should I use web site or website? Someone just suggested
swapping opinion for guesswork as it sounds better in the second
sentence? Maybe I should say text instead of graphic in the first
sentence? Should I use a two-column design for the page instead of
three, or split the text into two columns? Maybe I should change
the page round completely. So many choices…
The problem is that choices like these usually come down to
guesswork, personal preference or long-held opinion. Sometimes we
make the wrong choice for all sorts of reasons. Having lived with a
web page for a while, it can feel jaded and stale. There is a
natural urge to change it. To a new visitor to the page, the
existing page may actually be better than the new page put in its
place.
Marketing professionals use a variety of techniques to reduce
the guesswork. Google Website Optimiser puts two powerful
techniques to test variations of web pages to see which
variation performs best, into the hands of all website owners. To
do that you need to have a specific goal in mind for the page, such
as sign ups to a newsletter, adding a product into a shopping cart
or visiting a specific page. The page that you’re testing also
needs to have enough visitors for you to see which version works
best.
There are two types of testing you can do with Google Web
Optimiser:
- A/B Split Testing
- Multivariate Testing
A/B Split Testing
A/B Testing is the simpler kind of test. In an A/B test you see
how the original, “A”, page performs compared to variant, “B”,
pages. There can be more than one B page. A/B testing can be used
if the page is fairly low traffic (less than a thousand page views
each week). A/B testing allows the testing of pages in which one
simple change to the page is made, the overall look of the page
changes or parts of the page are moved around.
Multivariate Testing
Multivariate Testing tests different versions of a single page.
You can try out several different versions of the parts of the
page's text and graphics, to see which results in the most visitors
going to your goal. You can vary different parts of the page
separately and Google Web Optimiser will work out which combination
is the best. To use a multivariate test in a reasonable amount of
time, the page needs to have high traffic (over a thousand page
views each week).
Both kinds of test can be set up by someone with a bit of
technical knowledge by following the step by step process on the
Google Web Optimiser
website (opens new window). After there have been enough
visitors to the page, the tool can tell you which version
performs best. It will even knock out underperforming variations so
the testing won’t hurt you as it goes on.
This isn’t just a toy for techies though. It offers real
business benefits. It allows you to increase your conversion
rates, avoid changes that decrease conversions, and do
away with those meetings debating which version of a page is
best.