Split testing
- Thread starter sharkkiller
- Start date
Guides aren't too common (but some shitty ones probably exist), because what you need to split test is subjective based on your goals. Split testing keywords versus split testing landing page layouts have different test parameters but the concepts are the same, A/B testing, or using a multi value test (n>2) set with bayesian filtering. Also if you are working on SEO straight split testing may be a bit more ephemeral.
There is a decent book on LP optimization (with a nod to A/B testing) called "Landing Page Optimization". Amazon.com: Landing Page Optimization: The Definitive Guide to Testing and Tuning for Conversions (9780470174623): Tim Ash: Books
if you want to test different landing pages try this neat little php code (found it somewhere here on the forum and saved it, cant remember who wrote it though)
<?
$lps = array(
'http://127.0.0.1/landingpage1.php',
'http://127.0.0.1/landingpage2.php',
'http://127.0.0.1/landingpage3.php',
'http://127.0.0.1/landingpage4.php',
);
header('Location: ' . $lps[array_rand($lps)]);
?>
<?
$lps = array(
'http://127.0.0.1/landingpage1.php',
'http://127.0.0.1/landingpage2.php',
'http://127.0.0.1/landingpage3.php',
'http://127.0.0.1/landingpage4.php',
);
header('Location: ' . $lps[array_rand($lps)]);
?>
I don't know much about this, but I would have thought it's important to only test one variable at a time when you're starting out? If you change two things and you get an improvement, you don't know which one caused it.
So all you need is 1) two things where you previously had one, so it needs to 'split' somewhere between the two options. 2) only one variable changed between them, 3) a way to measure the performance of the two split options
So all you need is 1) two things where you previously had one, so it needs to 'split' somewhere between the two options. 2) only one variable changed between them, 3) a way to measure the performance of the two split options
How does that help you find out which one of two simultaneously changing variables caused a change?