Facebook Ads- Systematic Testing

So, I'm going to be running some facebook ads soon. I've done lots of training and have had some successful ads in the past. I wanted to start this thread to get some input on what you guys think would be the most efficient way to systematically test ad success.

Thoughts/Questions-
Should I run separate campaigns or just separate ads under one campaign?
I don't know if you have felt it, but sometimes I feel like certain ads, under the same campaign, seem to do much worse than I expect. I wonder if having different campaigns would change this. For experimental purposes I have ran the same exact ad, but one CPC and the other CPM, PCPM, etc. ANd guess what, the oen where Facebook gets the most $ seems to get shown the most, and the others barely get shown. Maybe it is because I am bidding against myself essentially and so the ones with the higher bids get more impressions? I've even tried running the same ad, but allocated different spending amounts to see which one yielded the best ROI.

So again, if you have thoughts about testing under a sigle campaign vs mutliple, I'd be glad to hear.

Otherwise, what would be good options to play with? I guess I am looking for technical things more than the ads themself. Assume I want to test 3 pics, and a few different calls to action. Should I try running one CPM and one OCPM? One CPC and one CPM? I was surprised to find out that on the last campaign I ran, the one set for CPC (in the newsfeed) was by far the most successful. I had like 30-40 shares which helped.

I feel like I get very opposite suggestions from all of the "experts" out there. Some people say to stay away from the newsfeed and only advertise on the right side. Others say that the right side is old news and the newsfeed is where it is at. I'll admit that with my last campaign, my best ad was in the newsfeed, it got commented, shared, and liked a lot (for me) and I also noticed that about 80% of the interaction was via mobile. I feel like there is still a market for cheap clicks with the smaller ads on the right, but with everyone going mobile, the right side ads are going to be a limiting factor on who you hit.


How long to "test" for?
How long do you guys let an ad run before pulling it? I have heard that it takes 2-3 days for an ad to get in the groove. Sound legit?

Test budget? How much do you guys budget for testing out new ads or ideas?

Anyway, all of your input is welcome. I love good conversation and hope to add and take away from it.