Why is it needed to test 2x-3x times the payout for each placement? (3)
Why is it needed to test 2x-3x times the payout for each placement?
Can someone explain what is the logic behind testing each placement to 2x-3x the payout and does it still apply if you're on a limited testing budget?
Also, how many landers do you usually test new offer with initially?
Lander wise, I usually test between 2 and 4 (usually 4) to begin with. 2 vastly different designs and copy so I can test copy1+design1, copy2+design1 etc.
Yes, the payout cut rule does still apply on a limited budget. Without a spend of 2-3+ times the payout, you don't have data that's even close to statistically significant... especially when you factor in the 3+ different angles you're probably testing at the same time. You could be cutting a placement that is going to really work for you. That's wasted money, both in testing cost and potential earning.
A single placement could make you anywhere between $x and $xxxx daily - you don't want to be making rash decisions. Someone else is gonna snap up that same (potentially amazing) spot, just because you didn't want to spend an extra few dollars to make sure.
Also when cutting stuff, keep in mind that conversions are often delayed - if you're doing apps, the user might not actually open and use the app until a day or two later. If you're on pins, the carriers probably won't process leads instantly. So give it a day or two. I finished running decent volume to an app almost a month ago now, yet still have leads trickling in today.
I started on a low budget too, so I know where you're coming from. Be smart, not cheap.
There are no strict rules you have to follow. They are just some guidelines to make it easier in the beginning.
(not just with this example, but generally. Not only in AM but in life 
)
In the end ask yourself what the purpose behind it is: To blacklist bad performing placements
Requires: Statistical significant data to prove a placement is bad
Q: What is statistical significant?
Q: Does it even matter if it's significant? The goal is maximum bottomline profit right? Maybe your profit will be higher if you cut earlier and avoid losses through many bad pubs? Maybe your profit will be less because you cut pubs too early?
Solution: Make a statement and then test it to find out what works best. You could split-test two different cutting strategies and see which makes the most bottom line profit in the long run for your specific setup.
That's basically how I approach such questions - and there are so many of them
For cutting, the smaller the impact of a placement, the earlier I tend to cut. If it's a huge volume pub I'd give it some more time to prove itself because the possible loss if it should be a wrong cut would be huge. For something like carrier or OS cutting, I'll usually go at least 20x - 100x payout before blacklisting them. Not only to know if "good" or "bad" but also to see exactly how good or bad for future options.
Also for a $0.30 app install I'd give several times payout to test, while for a $25 pin submit I would test 1x to 2x payout.
It also depends on the traffic source. If I got a lot of complete shit placements I blacklist earlier to avoid big losses. If I already got a global traffic source "shitlist" I can spend more without having to worry about big losses. Some pubs on some traff sources have specific indicators of proven non-performance (no lp click throughs) so you can blacklist them even at 0.5x payout and less.
You could also put "maybe" placements on a separate list, blacklist them early, and only keep running the really good ones. Then once everything else is optimized and the funnel is much stronger set the "maybes" to active again and give them some more spent to see if they might work.
Home > General > Affiliate Marketing Forum