Hi all,
I have invested pretty much money in EXO, but get grim results to now. As for the biggest problem I meet, I will give you the real briefing of an offer I have been running.
Vertic: Adult Subscription
conversion flow: Wap - Single opt-in
GEO:IN
Carrier: Idea
Payout: 0.42
I began to run the first testing campaign for this offer around May 2nd as is shown below. To cut petty nonsense, I will straightly keep to the point. If there is anything you need for your thought just ask plz.

As you can you see, this is a 3 days running result. the reason why zero conversion shown on the screenshot is I hadn't set up correctly with conversion postback, which was fixed latter though. And after 3 days testing, I got pretty promising, despite that its still red, statistical results as is shown below.

So I thought it's a nice offer and I can optimize to turn it green. some of you guys knowing EXo stuff have pretty good idea of how many sites or volume can it have. So I analyzed all sites in the testing campaign, and splittest according to the Impression volume(only those with large volume but have bad performance), winning sites, &bad sites.

After another 3 days running, I got the results above. The only campaign with low Ecpa is the last one, which is still not profiting. The rest of it either yields none or have unreasonable high costs.
But the the first launched testing campaign was running all the time to keep my winning sites list become more, during which I surly had excluded some poor performance sites(blacklists).
The thing is I encounter the same problem I had in a few testing, which is each time I add the winning sites to my whitelisted campaign I had bad conversion reports really a difference from its performance at the first testing campaign. Besides, the new sits just keeps coming to my first launched testing campaign. Theses new sites are not that costy respectively but as a whole it really increase my spending and lead me to unprofitable situation.
Sincerely expecting some idea here to help me out.
BTW, during the site spying analysis, I found that 300*250 size could be more promising when running adult offers on tier 3 geos. so I put it into test, but still I meet this problem again.
![Click image for larger version.
Name: ~YQ]OO[1V2I%FBU0M`_I127.png
Views: 15
Size: 38.9 KB
ID: 15366 Click image for larger version.
Name: ~YQ]OO[1V2I%FBU0M`_I127.png
Views: 15
Size: 38.9 KB
ID: 15366](https://i.imgur.com/VYT1Uum.png)
NEED your help badly NOW!!!