Home > The Newbie Zone > Questions and Answers

Facebook Testing Question (7)

12-03-2018 04:31 AM #1 nosyboy (Member)
Facebook Testing Question

Hey guys,

Just starting out with FB and was hoping someone could provide some insight to how they test. From what I have read, you should:

1 campaign -> 1-3 adsets(different) -> 1-3 ads(all identical, duplicated)

How would you test each variable? So assuming your offer is good (high epc), this leaves you with the following variables:

1. ad + placement + image
2. landing page
3. audience

My guess is your should focus on one of the above, and find the winner. Then test other variables, find the winner and move on

So an example would be:

Focus on the ad first (test out different images + text), keeping your audience and landing page constant. Then choose the ad with the highest ctr and test that ad to different audiences. Have an idea of what your target audience is and then testing multiple landing page to see which performs best.

Is this how I should approach it?

Thanks!

12-03-2018 08:35 AM #2 stickupkid (Senior Moderator)

You didn't read a lot based on your questions. So I suggest to read loads and loads more, but oke hereby a little bit of feedback how I test stuff;

Run 3 ad sets with each 3 different images (so 9 different images). Duplicating is only relevant if you know a certain creative has proven to be good.
Keep text, settings, targeting the same. So after some days you find the winner images, go test them with different angles/text and see which attract most attention (lowest cpc's).

Ask advertiser/network for look a like audiences OR ask them what the best targeting variables are for that campaign. Normally you already have like a rough estimation who your potentials are, but it really depends on your vertical. Running nutra or casino on a cold audience is very hard for example to make profitable.

Verticals like lead gen sweeps are quite easy to run without pixels or look a like audiences, most of time women 35+ will do the trick here. To give some more feedback let us what you are planning to test!

Not sure what you mean by landing page (offer page or prelander?)

Good luck!

12-03-2018 11:15 AM #3 nosyboy (Member)

Super helpful! I appreciate the suggestion.

I m going to start with lead gen sweeps because they seem the most feasible to do at this point.

I mean prelander.

I m just worried that facebook will ban me if I direct link offer to the ad. So if I do FB ads - > pretender (CTA) -> offer page, I should be okay?

12-09-2018 04:08 AM #4 nosyboy (Member)

Quote Originally Posted by stickupkid View Post
You didn't read a lot based on your questions. So I suggest to read loads and loads more, but oke hereby a little bit of feedback how I test stuff;

Run 3 ad sets with each 3 different images (so 9 different images). Duplicating is only relevant if you know a certain creative has proven to be good.
Keep text, settings, targeting the same. So after some days you find the winner images, go test them with different angles/text and see which attract most attention (lowest cpc's).

Ask advertiser/network for look a like audiences OR ask them what the best targeting variables are for that campaign. Normally you already have like a rough estimation who your potentials are, but it really depends on your vertical. Running nutra or casino on a cold audience is very hard for example to make profitable.

Verticals like lead gen sweeps are quite easy to run without pixels or look a like audiences, most of time women 35+ will do the trick here. To give some more feedback let us what you are planning to test!

Not sure what you mean by landing page (offer page or prelander?)

Good luck!
Stickupkid,

Quick question:

So let's say after testing the 9 images, I find the top 3.
Do I create an entirely new campaign, with 3 Adsets. In each adset, I will keep the image the same but with say 3-5 different text/angle?
Then find the winner again. Create a new campaign with say 3 adset (keeping the best image + angle constant) but with 3 different audience?

I split test with different audience to see which is the best converting?

Of course, during all this testing, I will have a pixel on my website to create a custom audience from it

12-09-2018 03:06 PM #5 Mr Baffoe (Veteran Member)

The Fb pixel event learning and optimization data is stored at the adset level.

You are essentially starting over when you create new adsets.

If you are not specifically testing targeting or bidding strategies, reuse the adsets you already have.

02-05-2019 09:54 PM #6 nayketjaem (Member)

Quote Originally Posted by Mr Baffoe View Post
The Fb pixel event learning and optimization data is stored at the adset level.

You are essentially starting over when you create new adsets.

If you are not specifically testing targeting or bidding strategies, reuse the adsets you already have.
If I want to test new creatives, do I just add those images to the winning adset? Because I heard making changes can also reset the learning phase

And when duplicating an adset, will the new adset keep all of the learning and optimisation days of the original adset ?


Sent from my iPhone using STM Forums mobile app

02-06-2019 05:34 AM #7 stickupkid (Senior Moderator)

Quote Originally Posted by nayketjaem View Post
If I want to test new creatives, do I just add those images to the winning adset? Because I heard making changes can also reset the learning phase

And when duplicating an adset, will the new adset keep all of the learning and optimisation days of the original adset ?


Sent from my iPhone using STM Forums mobile app
Test new creatives in a new campaign or new ad set, indeed avoid touching winning ad sets as much as possible.

New ad set means new learnings.


Sent from my iPhone using STM Forums mobile app

Home > The Newbie Zone > Questions and Answers