How much would you spend to test this properly (6)
5 Angles, 5 banners each - 25 banners
5 landing pages
Offer Payout is $3.75
depends on banner CTR
are all clicks being distributed evenly to the 5 landers or split up between angles?
if its a completely fresh test on a new offer/traffic source, i'd throw AT LEAST 2X offer payout at each variable. meaning - spend at least 2x the offer payout on each possible combination of banner > lander > offer scenario
Originally Posted by swiftclick
Is the offer proven? The LPs?
5 x 5 x 5 x ($3.75 x 2) = $937.5
If it were me, I'd start with 5 angles, 5 banners, 2 LP and would spend $20 on each angle overall just to get a feel for potential and then would expand from there.
Might as well get some preliminary aggregated data and use that as a benchmark for moving forward, rather than trying to go to statistical significance on every single combination off the bat.
Your weak point is having no basis at all. With dating and diet, I know the 2-3 landers that work best, and have an intuition about the ads already.
If you can instead test 3 angles and 3 landers (modelled after ones you've seen), you'd be better off IMO, then you can take the winner and split test against the 2 unused angles/landers. More efficient. Less money spent, and you can cut sooner and determine your control. Then keep testing and when you get to the stage where the control wins 8-9 tests in a row or only loses by <5%, you should stop and move on with expanding traffic sources and so on.
I'd do 5 angles, with 2 banners each, and 2 landers for each just to see if the angle is even decent.
The idea here is your narrowing down your $ for testing by finding which of the 5 angles are the best, THEN expand the winner with more banners/landers. I almost urge you to do this as finding the winning angles is the 1st goal.
Home > General > Affiliate Marketing Forum