• There's multiple ways to split test audiences, especially with the ability to use CBOs, or campaign budget optimized structures, so that you're able to test either one interest at a time, bundle of interest, or you can essentially create a cluster of your best performers so that you can have a campaign that's just producing for you at a very high volume in the future. 

• What we're going to break down first is how I go about split testing audiences, one by one, while still maintaining a positive return on ad spend. 

• What I typically do is I list out what interest it is that we're targeting so that we're aware of just exactly who it is we're targeting, why we're targeting it.

• In this example, this entire campaign is focused around the word superfood. So we're targeting the interest of superfood and we're trying to see just exactly how it performs with the copy and creative we're associating with it. 

• We have two variations here, a 1-2 and 2-2, essentially, it's the exact same video, but they've changed thumbnail and it has the same copy. So for example, the way that we do our naming convention, 1-2 means it's video one thumbnail. It's the same video. Number one is the first thumbnail. Number two is the version of copy. So two x two is the same video. Second version of the thumbnail, same exact copy. And each of these ad sets, which is all the same interest only has two variations within it. 

• Now I'm sure some of you were thinking like, why would you copy the same interest four times?  What we have found from our testing is essentially this is like a four by two method, four by two, meaning four copies or at least three copies in this case, and one of them being the, the main interest ad set and then duplicating it three additional times, you're creating four variations of ads. 

• That's what the exact same audience set up the exact same placement setup, same budget. Cause it's a campaign budget optimized campaign set to $30 a day and then 30 euros a day. And then we're going to have two versions of the ad within each one. So essentially four ad sets, eight ads total. And now what we're allowing Facebook do is to send the traffic to this specific interest group and then seeing which one of these ad sets actually deliver sales. 

• Now, what you might find interesting here, as you could see from it, a ROAS standpoint, all we're really trying to do is find the ROAS' of 2.0 and higher based upon this setup. So this is a four by two situation, we're doing four interests or four ad sets with two ads per interest group. Another one we've is four by four, which works really well. 

• If you were trying to do, let's say four different variations of video or four different variations of copy. If you're doing four variations of copy, you're going to use the same image or video creative. If you're doing four different videos, you're gonna use the same copy for all four. And once again, this is going to be dependent on the budget you set, but each of these are either 25, 130 euros a day. It could be in dollars as well. 

• We've typically done our testing at 25, 30, 50, 75, a hundred all the way up to even 500 and had this work. But we do the four by twos and the four by fours in order to give enough traffic to each of the ad SEG groups within the campaign, which has all the same interest.

• And then Facebook is going to optimize based on which one of these ad sets is producing the CPA that we're looking for. So in this case, we're looking actually for not just a low CPA below 20, we're actually looking for a row as above too. 

• So anything that's above a two, we keep on and sometimes you'll find that some of these campaigns, when you do this four by two or four by four method, you'll often find at least two of these start to perform and the other two don't. 

• So what he ended up doing is having each one of your ad sets begin to perform, or your campaigns begin to perform over time and you end up having, in this case, you could see here, we have nine different campaigns, different interest groups. Sometimes we'll even bundle together, all the interests. 

• We're looking to target into one centralized CBO and test each one of those individually before breaking them off.

• So these are the same exact interests that we've created individual campaigns, but you could see here, these are off. And these three are working at least in this specific example from a ROAS standpoint. So the same nine interests we're creating individual campaigns. 

• We also created a bundle campaign with all of them within it at a higher budget. So that's how we go about split testing our interests. We typically do once again, 25, 30, 50, 75 or a hundred when we're doing our testing, we'll do a four by two. 

• So four variations of the, of the interest group and then two ads within each interest group. Or you could do a four by four for interest in four variations of the ad, creative within the ad set group. And if you do that, if you're going to do a four by four, I suggest doing a hundred a day or higher.

• If you're going to do a four by two, you could start out much lower. You could do 25, 30, even 50, and then go from there. But once again, you want to make sure to try some variations where it's an individual interest in each campaign. 

• And then you want to do ones where you clustered them together. In this case, I call it interest segmented where you're putting all the interests within, and then you're able to split test and see which of those interests are actually going to get the best performance If bundled together versus if you're split testing individually.

• You might want to put a single interest into each campaign. This might be more labor intensive, but with the new CBO setups, this is how we're optimizing our campaigns, split testing audiences, and actually able to scale it counts pretty effectively with low budgets as well. So that's how we go about split testing audiences. And of course you can get very creative. They've doing four by twos or even four by fours. And if you want to take it further, you could bundle everything together and see what happens from there.

David (00:01):

Hey everyone. Welcome back in this video, we're going to be focusing on split testing audiences. Now there's multiple ways to split test audiences, especially with the ability to use CBOs or, or campaign budget optimized structures, so that you're able to test either one interest at a time bundle of interest, or you can essentially create a cluster of your best performers so that you can have a campaign that's just producing for you at a very high volume in the future. But what we're going to break down first is how I go about split testing audiences, one by one, while still maintaining a positive return on ad spend. So as you can see here in this campaign, what I typically do is I list out what interest is that we're targeting so that we're aware of just exactly who it is we're targeting, why we're targeting it.

David (00:51):

Yeah. The ad copy associated with it. So in this example, this entire campaign is focused around the word superfood. So we're targeting the interest of superfood and we're trying to see just exactly how it performs with the copy and creative though. We're associating with it. So we have two variations here, a one dash two and two dash two, essentially, it's the exact same video, but they've changed thumbnail and it has the same copy. So for example, the way that we do our naming convention, one dash two means it's video one thumbnail. It's the same video. Number one is the first thumbnail. Number two is the version of copy. So two by two is the same video. Second version of the thumbnail, same exact copy. And each of these ad sets, which is all the same interest only has two variations within it. Now I'm sure some of you were thinking like, why would you copy the same interest four times?

David (01:50):

Well, what we have found from our testing is essentially this is like a four by two method, four by two, meaning four copies or at least three copies in this case. And one of them being the, the main interest, uh, ad set and then duplicating it three additional times, you're creating four variations of ads. That's what the exact same, um, audience set up the exact same placement setup, same, same budget. Cause it's a campaign budget optimized campaign set to $30 a day and then 30 euros a day. And then we're going to have two versions of the ad within each one. So essentially four ad sets, eight ads total. And now what we're allowing Facebook do is to send the traffic to this specific interest group and then seeing which one of these ad sets actually deliver sales. Now, what you might find interesting here, as you could see from it, a ROAS standpoint, all we're really trying to do is fine.

David (02:46):

The [inaudible] of 2.0 and higher based upon this setup. So this is a four by two situation, we're doing four interests or four ad sets with two ads per interest group. Another one we've is four by four, which works really well. If you were trying to do, let's say four different variations of video or four different variations of copy. If you're doing four variations of copy, you're going to use the same image or video creative. If you're doing four different videos, you're gonna use the same copy for all four. And once again, this is going to be dependent on the budget you set, but each of these are either 25, 130 euros a day. It could be in dollars as well. We've typically done our testing at 25, 30, 50, 75, a hundred all the way up to even 500 and had this work. But we do the four by twos and the four by fours in order to give enough traffic to each of the ad SEG groups within the campaign, which has all the same interest.

David (03:44):

And then Facebook is going to optimize based on which one of these ad sets is producing the CPA that we're looking for. So in this case, we're looking actually for not just a low CPA below 20, we're actually looking for a row as above too. So anything that's above a two, we keep on and sometimes you'll find that some of these campaigns, when you do this four by two or four by four method, you'll often find at least two of these start to perform. And the other two don't. So what he ended up doing is having each one of your ad sets, uh, begin to perform, or your campaigns begin to perform over time and you end up having, in this case, you could see here, we have nine different campaigns, different interest groups. Sometimes we'll even bundle together, all the interests. We're looking to target into one centralized CBO and test each one of those individually before breaking them off.

David (04:34):

So these are the same exact interests that we've created individual campaigns, but you could see here, these are off. And these three are working at least in this specific example from a ROAS standpoint. So the same nine interests we're creating individual campaigns. We also created a bundle campaign with all of them within it at a higher budget. So that's how we go about split testing. Our interests. We typically do once again, 25, 30, 50, 75 or a hundred when we're doing our testing, we'll do a four by two. So four variations of the, of the interest group and then two ads within each interest group. Or you could do a four by four for interest in four variations of the ad, creative within the ad set group. And if you do that, if you're going to do a four by four, I suggest doing a hundred a day or higher.

David (05:23):

If you're going to do a four by two, you could start out much lower. You could do 25, 30, even 50, and then go from there. But once again, you want to make sure to try some variations where it's an individual interest in each campaign. And then you want to do ones where you clustered them together. In this case, I call it interest segmented where you're putting all the interests within, and then you're able to split test and see which of those interests are actually going to get the best performance. If bundled together. Versus if you're split testing individually, you might want to put a single interest into each campaign. This might be more labor intensive, but with the new CBO setups, this is how we're optimizing our campaigns, split testing audiences, and actually able to scale it counts pretty effectively with low budgets as well. So that's how we go about split testing audiences. And of course you can get very creative. They've doing four by twos or even four by fours. And if you want to take it further, you could bundle everything together and see what happens from there.