Website Optimizer from Google
Search marketing practitioners need their ads to be successful. To
help build that success, click-through and conversion tracking are
used to gain insight into user behaviour on their sites’ pages. If
visitors are reaching a site but do not take the action which is the
intended goal, what is the remedy? This is the time to improve the
site itself. This is where Website Optimiser can help.
The Website Optimiser allows online marketers to test changes in the
content of their Web sites’ pages in order to determine what will be
most effective in getting conversions. The marketer chooses which
parts of a page to test – for example, headline, image, promo text –
and sets up the tool to run experiments on a portion of the site’s
traffic to determine which content users respond to best. When the
software has collected enough data, it provides reliable,
statistically sound reports and a suggested course of action to
optimise the site for maximum conversion and business results.
Preparation
Before performing an experiment, the tester should take some time to
consider which aspects of the page should be tested and how; the
more thought put into the experiment, the more valuable are the
results which come out of it. In the support pages, Google offer a
Getting Started Checklist which gives a good idea of the steps
needed to prepare for an experiment. Google suggest that it might be
helpful to review the technical overview as well.
Running the tests
Running the experiment requires the assistance of the webmaster or
site administrator who inserts into the site the experiment’s code
(contained in the form of tags edited into the html) that enables
Website Optimiser to direct variable amounts of traffic to the
different versions of the page. When the pages have been tagged and
all the page and content variants have been created, Website
Optimizer can be started.
Multivariate and A/B testing
Website Optimiser uses multivariate testing, which is a method of
experimentation that allows testing multiple variables – in this
case, sections of a page – simultaneously. For example, using
Website Optimiser, the tester could identify the headline, image and
promo text as the page sections that he would like to test, and then
create three different variations for each page section. Users
visiting the page during the experiment might therefore see Headline
A, Image B and Promo Text C all together, or Headline B, Image C and
Promo Text A, or any other combination.
Website Optimizer can also be configured to perform simpler A/B
testing. An A/B experiment tests the performance of two (or more)
entirely different versions of a page. The experiment should start
with an original test page (the page whose content is the subject of
improvement) and alternative versions should be written. There are
no restrictions on how much the alternative test pages can vary from
the original. The tester can change the content of a page, alter the
look and feel or move around the layout. Website Optimizer will
direct traffic to the original page and to the alternative versions
to see which one users best respond to.
Use A/B testing when:
page traffic
is fairly low (less than approx 1,000 page views per
week)
the experiment
involves, changing layout, moving sections around or
changing the overall look or design of the page
Use multivariate testing when:
page traffic
is fairly high (more than approx 1,000 page views per
week)
the experiment
involves trying multiple content changes in different
parts of the page simultaneously while the overall layout stays the
same
Google's 'All you
need to know' about Website Optimizer
Website Optimizer
overview (Flash)
Website Optimizer
user guide (Flash)
