When you are talking PPC there are a few factors you need to look at - Obviously Return of Investment ( ROI ) is of big concern, but things like conversion ratios and and cost per conversion are huge factors as well.. I think the BIGGEST factor... Quality Score. A high Quality Score can REDUCE your PPC expense. I work around the concept there are 6 fragments in determining Quality Score they are:
1. Account Score
2. Keyword Level Score
3. Landing Page Quality Score
4. Mobile Score
5. Ad Group Score
6. Ad Level Quality Score
From experience, I can tell you, that you REALLY need to focus on this stuff. I would say the BIGGEST failure your average IM'er today has.. is Mobile Score... it sounds stupid.. but go look for yourself.. how many people actually have a mobile version of their landing page? ok ok you are saying.. well MOST pages are responsive..and I say ok great.. how many landing pages say "Insert e-mail here and d-load this file"? what will a mobile user do? They cant download the file right? they bounce... a simple "insert your e-mail here and we will e-mail the file to you right away!" becomes a mobile friendly landing page. And just because... If you think well I will just use the mobile language and everything will be fine... on a desktop they want it now.. they dont want to wait... that is why you need 2 separate landing pages
Aside from that..i pay a lot of attention to items 2 and 3. Keyword selection and landing page. The answer to nailing these down is testing.. the problem is to test this on the network you are wanting to advertise on.. well you will drop your Quality Score in the process. So you have 1 of a few options.. you can have ore than one PPC account... one to test with,and the other to run with over the long term to reduce your rates.. OR you can seek out additional traffic sources to test on.
I do what I call segment testing. Your standard PPC flow looks like ( Bait - Lander - Thank You ) Bait being the ad just in case. So the first thing I want to test is the bait. Depending on the type of ad I am running be it text or image ( banner ) I look for suitable places to test. i generally will use Facebook for graphic ads, and I use Twitter for headline text ads ( like Google Adwords ) At this point all I am looking for is what bait gets the most clicks to my landing / offer page. I could give a rats anything how that page is converting at this point. I simply want to find the headline or the graphic that is going to draw them like mesquitoes to a bug zapper!
Often.. especially with twitter in this process I am just throwing up text to get the click ratio. I just send them to my feed page and could care less about any conversion what so ever. THIS decreases overall testing cost.. same is true with FaceBook.. just drop them off to your timeline.. you can test for PENNIES!
Once you have that tested out and have some clear winners... you then want to develop the landing pages around those terms you used to get traffic. The message you use to get them there NEEDS to match once they get there. This is a part of how you develop a good Lander Page Quality Score. NOW, I focus on conversion, and from there I do what I need deeper in the funnel until i have a pretty decent converting machine, and all of the steps are hitting on all cylinders.
This is when I introduce the campaign over to Adwords or Bing etc. This process minimizes your testing expense and increase your overall Quality Score - because you are introducing a winner off the bat.
Hope that Helps!