CVR went down but users continue to click - MOBILE (9)
Hi guys i have start mobile and i got an strange case.
I splinted by 4 carriers
First days i spent some budget to test carriers and thought i had a promising segment for this campaign in Carrier named Claro:
Payout: 0.45
Traffic Source: Leadbolt
Carrier: CLARO
Banner 1
Clicks: 32
Conversions: 8
CVR: 25.00%
Banner 2
Clicks: 42
Conversions: 2
CVR 4.76%
So, all them was 3 times the payout. So, I thought banner 1 was great, since CVR was 25%.
But suddenly, banner 1 stopped converting all that much. Yesterday and today i've got
Banner 1
Clicks: 125
Conversions: 7
CVR: 5.60%
Banner 2
Clicks: 75
Conversions: 2
CVR 2.67%
Does anyone has ideas about it? At first i thought i had find gold, but then i got frustrated again with that campaign.
PS.: Leadbolt does not provide frequency cap 
To be honest those amounts of clicks are just not nearly enough to make any decisions on...especially on Leadbolt. I would say once you get 10x the traffic volume you can start making informed decisions.
You mean something like 1250 clicks deondup?
I thought before 125 clicks would be enough since that equals to $6,25 - 13x offer payout
Thx for help
This is a common misperception among affiliates.
Statistical significance has nothing to do with the
amount of any payout.
Got confused now.
Two tutorials point me here to spend 3x to 5x the offer payout and stick with the winners
May you help me being clear about that matter?
I used the Binomial Confidence Intervals, is about that are you saying? But the calculator returns me 11% minimal.
You are confusing two distinct concepts.
Statistical significance doesn't care about your payout. Statistical significance is a measure of probability. The probability of you winning the lottery with a single ticket doesn't change if the pot this week is $50 million instead of $25 million.
Your payout only effects your expected value. This is calculated by multiplying the array of possible outcome probabilities with their associated payoffs.
Really thanks cmdeal. Got you about the payout concept.
And about Statistical significance, why had deondup said i need 10x more? How did he reach at such value?
As cmdeal says, statistical significance doesn't care about payout. It only cares about the number of trials (impressions) and the number of successes (conversions).
The reason that payout is frequently conflated with statistical significance is that, given the wide variance in costs of data, a payout multiple is a useful shorthand for the amount of data you'll usually need to acquire for statistical significance assuming you expect ROI to be within a specific range.
If you are making some assumptions on the range of ROI you're seeking in a campaign (essentially the usual range of ROI for an AM campaign) and the approximate payout range (again, standard ranges for a CPA campaign) you can derive some rule-of-thumb calculations based on payout for how much traffic / data you need.
That's why we have rules like 4x payout for an initial campaign test on POF, for example.
If you unfortunately read what jthomas said, try to strike it from your mind.
Home > Mobile > Follow-along Campaigns