FEATURE
great way to do this is by mapping each
page in the funnel. Each page should
have a list in each of the following:
Essentials: a list of all essential steps
the user needs to perform in order to
successfully move to the next stage. For
example: filling out form fields, clicking a
button, etc.
Focus elements: page elements you want
the user to focus on in each page. For
example: offer, pricing, buttons, etc.
Distractors: elements in the page that
might distract the user’s attention. For
example: a menu, images, etc.
After you complete this mapping of
your funnel, it might be a good idea to
look around at similar sites and do exactly
the same. This will help understanding
the benchmark in the industry. Now,
looking at the mapping of your own
funnel, and using your experience from
watching others complete it, you should
have a better understanding of what to
test. Basically, it’s usually a matter of
decreasing the number of essential steps
and distractors, and making sure the focus
elements get the ‘respect’ they requi re.
Still, there’s no best practice here, the best
practice is to continue testing.
Some funnels need to be long in
order to help the user feel more confident
with the service; some users need less
of a funnel.
68
iGB Affiliate FEBRUARY/MARCH 2014
Five testing guidelines
4. Clean-up after testing
In order to perform a test that will end
with meaningful results, you should always
follow certain guidelines.
After a test is completed, it’s important to
pick the winner, remove the testing tags
and redirect from the less performing
pages to the winning ones in case one of
the users bookmarked any of them.
1. Some tests are just too small
It is important not to test small changes;
you will often find that the test results aren’t
significant enough. There are known cases
where conversion rates have significantly
improved by the mere changing of a button
colour, or a call-to-action test. These are
usually free products or trials; when it comes
to something more complex, it’s quite rare
that it will yield a meaningful improvement.
5. Repeat
Testing isn’t something you do once in a
while, and it certainly isn’t something you
only carry out when you have a problem.
Testing should always be performed on
your site, and when you find a winner, you
should start thinking of the next test using
the methodology we have described in this
article, and follow the same guidelines.
2. Don’t test for too long
Certain periodical effects could be messing
with your results. Automated testing tools
use randomisation to cope with these
unknown factors, but often, even this is
insufficient. You should test for the shortest
period of time possible.
3. You can’t copy and paste
If, in one of your sites, you broke down
the registration form to three different
pages, it doesn’t necessarily mean it will
work on any other site or with a different
kind of traffic. Testing results are only valid
in the specific testing environment. They
can only give you an idea on what to test
somewhere else.
ELI UZAN is Head of eGaming
at Google Israel. Throughout
his time at Google, he has
been providing consultancy
services to large iGaming brands in
different areas of online marketing,
business development and product.
Before joining Google, Eli was the
co-founder and COO of an Internet
company. Eli holds an MBA and a BSc
in Computer Science from the Hebrew
University of Jerusalem.