Shopify Ecosystem

Audience Split Testing

audience-split-testing

Split testing audiences can be a challenge.

We asked the senior media buyers at Pilothouse three big mistakes brands MUST avoid when testing audiences.

Here’s what they said. 👇

🚨 ABO not CBO:

ABO = Ad Set Budget Optimization (you tell Facebook to spend X amount for every ad set in a campaign).

CBO = Campaign Budget Optimization (you tell Facebook to spend X amount on this campaign and Facebook can decide which ad sets to spend your budget on).

The best way to split test audiences is to have one campaign wherein each ad set is a new audience that you want to test with its own budget.

This will allow for even spending, so that you can compare your audiences and their data – apples to apples.

CBO campaigns will push most of the budget into whichever audience FB thinks has the most potential.

Usually the key is to take these winners from ABO tests and look for ways to scale them out into a CBO campaign.

👊 Don’t Stack Too Many Audiences:

If you’re running very large broad audiences when testing, don’t stack them unless they’re all similar in interests or intent.

Keep audiences separate so you can see from your results which audience was driving your results.

If you have a smaller audience you may need to stack to get the audience size you need – that’s okay.

Just make sure that your scale campaigns also have the same audiences stacked.

🦾 Detailed Targeting Expansion turned off:

At the ad set level you can select “Reach people beyond your detailed targeting selections when it’s likely to improve performance.”

In scale campaigns this should always be selected, but for testing purposes you need to leave this turned off.

While testing you’re looking for data, not purchases, and if Facebook is going outside of your selected audience then you’re not getting clean data.

You may think that audience is great and move it into scale campaigns, when in reality it was users outside of that audience that were driving the data you saw.  

Look for early indicators of success and then move those audiences into your scale campaigns.

Special thanks to our friends at Pilothouse for their insights on this topic.
I'm also on