by Mark Joyner
Marketing should be treated like a science. If you are serious
about making money on the Internet, it is absolutely crucial that
you spend some time testing your results and refining your approach.
Let me drive this point home. Say for example you have a website
that gives you a visit to sale ratio of about 200 to 1. Not really
bad as far as web sites go. Most do far worse.
Now, assume that you get 5,000 visits per month. Do the math.
That’s 25 sales per month. If your profit from each sale
is $50, you are making $1250 per month profit from your website.Now,
what would happen, if by changing one small thing on your website,
you could improve that ratio to say 150 to 1. It may not seem
significant, but let’s see how it works on
Now, instead of 25 sales per month, you are doing 33. Multiply
that by $50 and you are now making $1650 per month. That’s
a difference of $400. It’s not like you’re having
to work any harder for that additional $400. Your website is always
Now, the example I have shown you here is quite mild to some
of the drastic improvements I have seen as a result of market
testing. All my life I’ve heard people say "don’t
work hard - work smart!"
This is excellent advice. Working smart means getting more from
Well, testing is a systematic way of helping you to work smart.
It’s just like the scientific method, really. You may remember
it from school. There are many versions of the scientific method
for different disciplines and many scientists will argue about
which approach is best. For our purposes, let’s take a very
bare-boned and simple version. It’s really all we need:
- Generate a hypothesis.
- Test your hypothesis by performing an experiment and recording
- Perform calculations and draw conclusions.
- Confirm or refine your original hypothesis.
For example, let’s say in the above example, we hypothesize
that by adding a money back guarantee, we will get a better visit
to sale ratio. This is our hypothesis. To test this, we make the
change on our websites and record our results. That’s our
experiment. After a few weeks (or any period of time we determine
to be sufficient) we conclude the experiment and perform calculation
on our data. We then discover that the visit to sale ratio did,
indeed, improve during the course of our test run. We can then
draw a conclusion that our original hypothesis is correct.Seems
pretty simple, right? Well, actually no...
*Controls and Statistical Significance*
In the preceding example, we did a pretty straightforward and
simple test. However, the results we have drawn may not be valid.
For example, what if the majority of visitors we received before
the test run came as the result of a banner ad campaign we were
running? Then, without us knowing it, sometime during the campaign
a company favorably reviews our product and drives people to our
site. It could be the case that people who read that review were
more likely to buy the product.
It could also be the case that they were much more likely to
buy and that adding the guarantee lessened the effectiveness of
the site. Or, it could even be the case that the guarantee did
increase the effectiveness of the site, but not as much as we
The problem is, we just don’t know.To address this issue
and to improve the reliability of our tests, we need to establish
"controls". A control is a measure that allows us to
isolate the factors which are causing the effects we record.
For example, a better way to run this test would have been to
run two websites at the same time. Both of them identical in every
way (promoted the same way, designed the same way- everything)
except for the one element we think will make a difference. The
site without the change is called the control site. The site with
the change is the test subject.
This same principle can be applied to any other type of marketing.
If you want to test the effectiveness of a banner ad, run two
banner ads on the same site for the same period of time. If you
run one banner ad on one site and the other on yet a different
site, you can’t be sure that your results are valid.
Using strict control techniques improves the chances that your
findings will be valid.
Even if we use good controls, we can still never be sure of our
results. There may be something affecting the test we haven’t
thought of. That’s why we also need to resolve ourselves
to the fact that the results of one single test should not be
accepted as conclusive.
Rather, we need to perform a great many tests and look at the
over all trends. Only after we have compiled a considerable amount
of information can we safely draw any conclusions.
Just how much testing is enough is up to you. You should test
until you are satisfied, but be honest with yourself. Don’t
just stop testing out of laziness!
*Turning tests into profits*
The idea here is to test as many different methods as possible,
discover the most effective method, and crank up the volume on
your best method. When you get into serious volume and your business
is really doing well, even minor differences in results will have
a major impact on the amount of money you take home. And sometimes
very subtle and minor changes to a website will make huge differences
in your results.
If you could find out which of your advertising campaigns was
bringing in the most money for you, wouldn’t it make sense
to put more effort into that campaign? Or do more of that type
of campaign? Wouldn’t you be crazy not to?
In the following chapters...
... we’ll show you exactly how to track and monitor the
results of all of your Internet marketing efforts. After we show
you the logic behind each method of testing, we’ll show
you the physical tools and techniques you need to employ to track
Tracking is just recording the results of one campaign or another.
Testing is the systematic monitoring of these campaigns to discover
what is working and what is not using the scientific method.
Article by Mark
Joyner, CEO of Aesop Marketing Corporation and creator
Killer Internet Marketing Tactics" - a *must have* tool
for anyone serious about doing business on the Internet. Do yourself
a favor and check this one out today!
here to send this article to a friend...
Back to Articles Page