I was running this campaign for Brazil on Xhamster 300x250. 20 Variations (10 banners x 2 LPs).
Offer payout was 0.90/SOI
I sent around 1x the payout for each variation:
"Variation A", 1 out of that 20 took me 3 conversions:
0.90 spent for 2.70 return;
Then I stoped the other 19 variations and ran "Variation A" alone, with 3x/24h frequency cap.
I then sent 20 USD worth of traffic. However I got to see just 10 USD of return.
########
Other time the same thing with this US targeted campaign. Xhamster 300x250:
20 variations >>> 1x payout spent on each
1 variation brings 3 conversions, like this:
I spent $ 5 on it and it made me $ 15
I stop all the 19 and just ran this 1 winner alone, with frequency cap.
Then I spent $ 20 on it and it gave me only 2 conversions worth just $ 10.
So my question is: why that happens? How to make the "winner" keep that same great ROI it once gave? Or even better, how to find a REAL winner?
Xhamster is a tricky traffic source. Run your banners on other exoclick premium sources and you will find better converting source.
No sample size at all. 1x the offer for creative & LP combo is by far not enough to make any judgements in my opinion. No matter you're working with a $1 or $30 payout, spending just the payout for a 'test' can be seen as more of an indicator than a real test. Nothing extraordinary for this to happen.
If all banners been a totally different angle/layout, try duplicating those. Same for the lander. Run a test again with at least 3x the payout spent per creative & LP & offer combo and make your decisions based from there.
^^ Exactly.
Let's say that you're paying $1 CPM for a placement (high but possible) and you've got a payout of $2. You run banners for 1x the payout (2000 impressions), and Banner 1 converts, whilst Banner 2 doesn't.
That means Banner 1 is a winner while Banner 2 is a loser, right?
Not so fast. Let's look at the statistics using the usual Statistical Significance Calculator.
Banner 1, which converted, has a predicted conversion rate between 0.001% (or therabouts) and 0.28%. That means that over a million impressions, it could make you nearly 200% ROI, or it could lose all your money, or somewhere in between. You don't know yet - more testing needed to be sure.
And Banner 2, which didn't convert, has a predicted conversion rate between 0.0% and 0.18%. Which means that over a million impressions, it could lose all your money, it could make you nearly 100% ROI, or anywhere in between. Again, you don't know yet - more testing needed to be sure.