Having multiple targeting options at your disposal is always a good thing, but sometimes you need to know which one works best.

Enter: the split test.

Just because one audience works in one context doesn’t mean a different one won’t be better in another.

And in most ad campaigns, starting with more than one option is the best way to go about it.

Fortunately, there’s a simple way to test multiple audiences at once, no matter the campaign type or your advertising goals on the other side of it.

Create a new campaign

To implement our split test, the first thing we’re going to do is create a new campaign.

Audience targeting happens at the ad set level inside Meta’s ads manager, so our end goal is going to be to pit two ad sets against one another inside of this single campaign and ultimately narrow that down to one.

For this example, we’ll create an engagement campaign, but this methodology will work with any campaign type you choose.

At the campaign level, the only setting that needs specific attention here is how we apply our budget.

The effort is to allow Meta’s system to allocate our ad spend to the better of the two ad sets (we’ll use that information to determine our winner), so we’re going to leave Advantage campaign budget set to the “On” position and set our daily ad spend here at the campaign level.

Moving on to the ad set level, we’re going to set all of this just as we would with any normal campaign.

Because this is a conversion campaign, we’ll set the conversion location to our website and select our conversion event of choice (in this case, View Content).

Moving down to the Audience section, we’ll adjust our countries and age breakdown to match our preference, then update our detailed targeting to match.

In this case, I’m going to use a lookalike audience and only focus my attention on the United States, where I live.

Our final step at the ad set level is to identify where we want our ads to be shown.

For this campaign, I’ll select Instagram Reels and Stories only.

Now, moving on to the ad level, we just need to create an ad.

For this example, I’ll upload a video for my song, “To The Wolves”, adjust the text and CTA, include a link to my landing page, and we’re all set.

​

Duplicate the ad set

So now that we have our initial campaign set up, it’s time to introduce our split test option.

To do that, we’ll click the three dots to the right of our ad set name and select “Quick duplicate”.

Once our new ad set is loaded, we want to scroll down to the Audience section again to adjust our targeting options.

The point of this is to isolate the one variable we want to test (our audience targeting), so we’re going to make changes here and leave everything else the same.

Now, because I’m using a lookalike audience in my first ad set, I’m going to remove the lookalike audience here and opt for Spotify-only targeting in this second ad set.

Once this edit has been made, all that’s left is to hit publish and let the system run for as long as it takes for a clear winner to emerge.

​

Analyze the results

Once we have enough data in this campaign, our end goal is to turn off our losing ad set and keep to the single winning option.

Depending on ad spend, this could take days or weeks to accomplish, so check your ads regularly, and when you feel you have enough information to make an informed decision, go ahead and turn off that losing ad set.

As an example of what this looks like, I recently tested two ad sets with a similar setup to the example above, allowing Meta’s ad platform to choose between lookalike targeting in the first ad set and more detailed targeting options in the second.

In this example, the results were quite clear.

The lookalike ad set had a lower cost per result, more results overall, and a higher click-through rate.

Turning off the targeted ad set was an easy decision, and it cost me less than $150 (much less, in fact) to figure this out.

So now, with this one ad set running, I’m free to move forward with split testing additional ad creative at the ad level now that I’ve got my audience targeting dialed in.

But what if the results aren’t always this clear?

Then what?

Well, I think of it in terms of hierarchy.

The first metric I’m looking at is cost per result; however, if there’s no clear winner there, I’m looking at the number of overall results.

If a clear winner still has yet to emerge, the next metric I look at is click-through rate.

After that, if I’m still not sure enough to make a decision, I will look at cost per click and, finally, CPM.

And if, after all of that, I’m not seeing enough data to turn off an ad set, I’ll continue to let the split test ride until I have enough information to make adjustments.

Given enough time, a winner will emerge, and it will be time to start split-testing ad creative instead.

​