whenever I start a new campaign in a new geo, one of the first things I will do is split test the offers on 1 network, then split test that exact same offer between multiple networks. Now, when I do that its pretty common to see a significant difference in EPC regardless of payout.
I will generally split test between adsimilis, MUNDOmedia, and ClickDealer who all have a very strong reputation. yet between those networks, I can generally expect to see a EPC of almost double, regardless of the payout.
Right now, I am split testing the same offer between Adsimilis and ClickDealer. despite Clickdealer having a larger payout, right now I have Adsimilis with a EPC of $0.22, and ClickDealer at $0.07 EPC with just over 500 clicks split between them.
now I don't really consider 500 clicks to be a very significant amount of data with the offer at $2.xx, but its a pretty strong indication of what to expect.
I can understand that it is quite common for the advertiser to shave leads, but the difference between the networks always tends to be quite significant.
what exactly causes this?
cheers
Clickdealer in your example i'd guess either have way worse loading times or they are scrubbing. Although that is a very small amount of clicks. 250 clicks and to an offer would be only 25 leads with a 10% cvr... But if your cvr is around 5-7% then there can't be many leads seperating them?
I would say test for a lead amount of 50 or 100.
500 clicks (if it's TO THE NETWORK) is bordering on statistically significant data I think. Make the switch. FWIW, I've found CD to be excellent and have tested CD-Adsi in the past and CD has always won out in web dating. I suspect you might be doing mobile - maybe Adsi's publishers have a history of sending quality traffic to those offers, while CD's pubs are on initial probation with this advertiser.
Depends on where you're getting your EPC data from as well. Different tracking systems track a click different ways. It's always best if your'e split testing between networks to calculate on total revenue per network / total clicks served by traffic source or you tracker. CAKE, tracks clicks differently then Hitpath, then Has Offers. Then you take someone like Mundo with an unregulated 3rd party private system and you don't know what games are being played with click scrubbing to make epc's look better for the lambs that just use system stats. Another thing we see a lot is literally just tracking issues. 50% less = http pixel being on an https offer to me. Or multiple thank you pages and they forgot to put the pixel on one of them.
Another thing you need to consider. You can't run a split test for a few days typically if you've been running consistently with one network. You're going to get a lot of cookie sales that come in over time. So if you split in another network for a day they're going to lose almost every time unless network 1 is doing some super shady shit. Also depend on volume if you were running 3k leads a day 2 weeks ago and now 200-300/day split between two network... yes we see this cause people think things went wrong and their epc's are way down. Then they go looking to see if the network is the problem at that point. But Network 1 would have a ton of sales coming through on cookies. Now the volume has dropped so as a % of total sales it could be huge.