Definition: A/B testing is a simple way to
sell more from your website by comparing
how standard website elements perform against alternate elements.
A/B Spilt Testing is a proven method used by
direct mail advertisers for decades. In their case for eg, they
might want to mail out 1 million sales letters.
What they initially do however is send out 1,000 letters.
500 letters say X. The other 500
letters say Y.
Depending on whether X or Y sells more, the advertiser now knows
which sales letter to send to the other 999,000 prospects.
The problem with A/B split testing is that it's linear so it doesn't
test combinations of variables. That's where Multivariate testing
is useful.
Multivariate Testing (MVT) is where you can isolate
the impact of each webpage element individually. Crucially, multivariate
tests can predict the best combination of website elements even
if you've never tested them. This allows you to get real results
very quickly.
Th Google Website Optimizer is great for MVT on small-medium sized
websites.
The Taguchi method used by Vertser and Offermatica
allows marketeers to make fairly accurate predictions based on
scientific analysis of the probabilities.
This Enterprise-level software for corporations effectively "guesses"
the best combination of website elements
for you to use simultaneously to help you maximise sales.
|
