Which Variable To Optimize First? (8)
In optimizing campaigns, there are many variables for us to test,
Offer
Lander
Bidding (Low Mid High)
Frequency
Cutting Placements
OS
Carrier/Wifi
Dayparting
Etc...
Assuming I'm running a new campaign, and I got my first round of data
Usually I will run campaigns and split test the major variables first
1. Offers
2. Landers
What should be up next to split test and optimize?
From my understanding, the next few major variables are:
- OS + Carrier/Wifi
- Bidding
Questions:
1) Are there any order of most important variables to test to the variables to be tested the last?
2) For OS + Carrier/Wifi and Bidding, should I test these 2 variables first BEFORE the offer and lander? I saw some Follow Alongs of people that test their Bids (High Medium Low) first
I normally go with the traffic flow - meaning it all starts with the traffic source.
Depending on your campaign settings - multiple geos etc..., number of landing pages you are pushing.
1. I would try to identify and get rid the low engaging subid's (you need a web analytics on your LP's, where you can identify with utm parameters the subid's) that have the highest bounce rate (this can be done in conjonction with your tracking provider).
2. Then I would identify the highly engaging LP's / the one that have the highest CTR
3. Granting you already had some conversions, I would get rid of the low converting offers - unless they have a specific characteristic related to useragents - carriers, ip's, devices etc..
4. Scale up traffic to the right LP's / offers to make sure previous steps are consistent
5. Adjust the bids according to optimization
6. Add a new LP variant with 10-20% traffic implementing what you've learned from 1,2,3
7. Repeat the all process all over again.
Hope that helped.
Everyone has a different system, but there are some general rules that you should follow.
The first thing you need to sort out is to find a good performing offer, because when an offer doesn't convert, you can have the best creatives or LPs but it won't perform well.
In my opinion, banners/LPs should come next, because again, if you are using poor banners or LPs, it's pointless to try to optimize any traffic source related variables.
Once you have this sorted out, you can start cutting under-performing targets, play with the bids etc ... To make sure you're paying as low for a specific traffic type as possible, you should make separate campaigns for the major targets like wifi/carrier or iOS/Android ...
On Facebook for me its 1) Image 2) Angle
when testing images, we start with 10 to 20, then look for image with highest CTRs. We pick 3 of the highest CTR images, after 3 days. next, we will test 10 angles.
So we setup 30 adsets total. 10 angles, for each image.
What that means is you ONLY change the text in each ad, leave the same image, and rinse and repeat.
Then we run this for 3 days, and look which one gets the lowest CPA.
We kill all others that are not breaking even, and take the top 2 that perform the best (most net profit) and duplicate the adset at least x3 each to scale. We also scale by duplicating and changing just 1 setting, I wrote a post about how to scale, you can find it if you search for my recent posts 
But there's like 31 ways to scale a winning adset, which I didn't post publicly.
On Google,
We take 5 banner ads. 10 angles.
We setup 10 campaigns, 1 ad group, and in it 5 banner ads. (The banner ads are always the same, they dont change only the TEXT aka angle on them changes in each campaign)
We then run this for 24 hours on high CPC (like $1) at offer payout x3 per campaign. So if your payout is $10, then its $30 per campaign - the budget.
Then we pick the one with most net profit, adwords automatically will optimize the best image from the set of 5 in the adgroup, unless you manually change that default setting to NOT do that.
GOOD LUCK YALL!
Thank you everyone for your input!
Is carrier & wifi considered a major variable to split test at the beginning of a campaign?
Example: I have 3 offers to test first, each offer has 2 proven LPs (that's total 6 campaigns)
After that (for the first few rounds of optimization to find the best offer first), is it recommended to separate into carrier wifi and test too?
Or I can test carrier and wifi AFTER I'm done with offers and landers?
The reason i am asking I read that some offers do better for wifi, and some do better for carrier
You are right, some offers work better on wifi and some on carrier, you must always spilt test the offers at first, since you dont know where they'll work
My system is: Spit testing the Offers -> Lander -> Traffic source -> Bid -> and then optimizing (cutting placements, carrier/Wifi, etc)
As others said - your first priority would be to mass test offers and locate well converting ones. LPs come after, you need proven landers too. Everything else at the traffic source should be the third layer.
In other words, if you have proper offers and landers, optimising the traffic sources shouldn't be that hard. Focus on testing offers instead of trying to dig into every little detail on a traffic source because you may often times end up with just a tiny pocket of profitable traffic which probably wouldn't be worth the time and effort you've invested into this one campaign.
The reason i am asking I read that some offers do better for wifi, and some do better for carrier
Everyone has a different method, maybe launch a whole bunch of campaigns and just try different ways of optimising in order to find out what's the easiest for you.
Originally Posted by aloeveraa1491
Whenever possible, I always test wifi in a separate campaign and carriers in another - the best is to have 1 campaign for each carrier. This also depends on the source, on some sources you can run a mix as the bids are not that different, on some it's not possible at all.
Home > Paid Traffic Sources > Mobile