Facebook PPC Questions

Falian

Falian

Banned
Nov 2, 2007
1,317
46
0
xrumerblast.org
Hey guys,

I'm about to make my first serious leap of faith into PPC and i've decided to go with FB. Before I start, I have a few split testing questions to ask yall.

Right now, I see three general ways that you can split test on FB itself (not incl. landing pages):

1. Ad Copy
2. Images
3. Demographics

Here's my current understanding of how I should go about split testing for 1 & 2:

1. Split test multiple Ad Copy versions. I keep the ad copy with the highest EPC and axe the rest. Rinse & Repeat.
2. Split test multiple Images. I keep the image with the highest EPC and axe the rest. Rinse & Repeat.

I'm having issues determining the most effective way to split test demographics. My initial thought was to do A/B testing using the "catchall net" mentality:

3. Start with a very broad demographic. Test a more narrowed demographic against the broad demo. If the narrow demo has a higher EPC, then that becomes the new "broad demo". If not, rinse and repeat.

If I have any logic flaws, then please point them out. Right now, I can't seem to grasp my mind around how to split test the right way.

For example, Can I split test 1, 2, & 3 at the same time? Or can I only split test one variable at a time, keeping the rest constant?

I see issues doing it either way. If I test one variable at a time, I could see this issue arising:

Ad Copy: A v B
Image: A
Demo: A

Ad Copy B has a greater EPC.

AD Copy: B
Image: A V B
Demo: A

Image B has a greater EPC.

Final Result:

Ad Copy: B
Image: B
Demo: A

Isn't it posisble that Image B and Ad Copy A could have had a greater EPC than the final result?

This issue also arises when I think about split testing Demographics, for example:

Gender: All
Age: All

Gender: M v F

Male has a greater EPC.

Gender: M
Age: 18-24 vs 24+

18-24 has a greater EPC.

Result:

Gender: M
Age: 18-24

Isn't it possible that Gender: F and Age: 24+ could have a higher EPC but we were never able to test it?


Also, I'm sure this is a dumb question, but just to make sure: If you have a good idea of who is going to be interested in your product/offer, then is it ok to narrow down the demographic somewhat yourself?

Lastly, I can get the CTR for each ad via Facebook, but I've heard that sometimes a lower CTR and higher conversion rate can result in a higher ROI. Facebook has a conversion tracker, but i'd prefer to keep that data out of their hands.

Should I use Prosper202 for this? If so, does each ad need a seperate tracking code?


Thanks in advance for the help and sorry for the long one!

- Nick
 


Hey it seems like you have a pretty good idea of what to do. As far as testing ads what I usually do is split test images first, find the ones w/ the highest CTR then start split testing ad copies/titles with that image.

As far as your concern about a different demographic might be converting better than another, there is no real way of knowing you just need to test, test, and test some more so you can find out.

Finally ya use prosper you don't need to make a seperate ad code for each. Just make an ad for your campaign then instead of doing the {keyword} at the end just put your ad name. So I do it like this ?sub=t1i1 for title 1 image 1 in my ads. Then just go to keyword tracking to see which ads are winning.
 
  • Like
Reactions: pegasus2218
cheezile said:
Finally ya use prosper you don't need to make a seperate ad code for each. Just make an ad for your campaign then instead of doing the {keyword} at the end just put your ad name. So I do it like this ?sub=t1i1 for title 1 image 1 in my ads. Then just go to keyword tracking to see which ads are winning.
Click to expand...

I just did a major facepalm. +rep for inadvertently pointing out my stupidity in making separate tracking URLs for each ad.

*sigh*
 
  • Like
Reactions: cheezile
cheezile said:
Hey it seems like you have a pretty good idea of what to do. As far as testing ads what I usually do is split test images first, find the ones w/ the highest CTR then start split testing ad copies/titles with that image.

As far as your concern about a different demographic might be converting better than another, there is no real way of knowing you just need to test, test, and test some more so you can find out.

Finally ya use prosper you don't need to make a seperate ad code for each. Just make an ad for your campaign then instead of doing the {keyword} at the end just put your ad name. So I do it like this ?sub=t1i1 for title 1 image 1 in my ads. Then just go to keyword tracking to see which ads are winning.
Click to expand...

I appreciate the advice man! I heard you need a hefty BR (> the 1k I have) to start PPC. So, I'll have to toss all this good info in the back of my brain for a while.
 
Even better way is the c1 c2 c3 c4 feature on prosper. So c1 = ad image, c2 = ad copy, c3 = gender, c4 = age...

Then you can track via the variable instead by one variation.
 
Bank roll.

@OP: Your ideas are fairly sound, but your means of reaching a conclusion from your tests is not. (As you clearly pointed out yourself.)

To draw conclusions you need to have valid results for all tests. Running a test against all demos then viewing males of 18-24 as having best EPC is flawed because if you have budgetary constraints then depending on the time of day, you might not even have your ad being eyeballed by certain demographics.

Also when you test multiple variables at once, testing time increases dramatically, therefore the financial commitment to reaching a conclusion is also stretched. You can draw conclusions quicker, cheaper and more accurately if you test one thing at a time.

I would suggest you just get stuck in, because as a great man said, "some people rehearse and rehearse and rehearse, but the curtain never goes up on the show."

You ain't gonna get everything perfect and the more you go over the perfect testing scenarios with a fine tooth comb in your head, the more daunting it will all seem. Chuck some stuff up, have a few ideas in mind and have a little fiddle. You'll be surprised how quickly the spending of money forces you to think and act with more clarity - well it does with me anyway!
 
No idea I'm afraid, although I fucking love that sales page. Absolutely fucking love it. Sales pages never make me stick, but that one did for a reason I now have to try myself. :D

It's a $4.95 trial for 14 days, and then only $49.99 so seems reasonable, assumming it actually teaches anything. Personally I dislike Gurushit a lot. Most of it results in, not always intentionally, paralysis through info overload.

By the way don't assume me to be an expert either. I've simply spouted off what I know, or at least believe to know! (Same goes for Mr Volks.) ;)
 
We're doing some Facebook PPC, here's 2 tips:

1. Facebook Ads Manager, like AdWords Editor for Facebook, makes split testing much faster.

2. In terms of Bank Roll it seems that initially, Facebook limits your budget to $50/day. Once you pay for a while, they'll boost your daily budget to $250 and then go up from there.
 
You must log in or register to reply here.
Top Bottom