One of the best ways to reduce costs and increase sales with Facebook ads is using split testing.
Unfortunately, splits tests with Facebook ads isn’t easy.
A key requirement of a valid split test is that audiences are distinct and do not overlap. But that isn’t possible with Facebook ad.
For example, let’s say you planned to test Optimization Delivery and wanted to see if Conversions and Link Clicks increased results. To do this properly, you’d have to target the same audience. The problem is that there is no way to split that audience so that 50% of that audience doesn’t see the same ad.
But it looks like Facebook has a solution – it’s called the Split Testing, and it’s just starting to roll out to advertisers.
How Split Tests Work
Split Testing is Facebook’s solution to run effective A/B tests on it’s platform.
There are three main features to Facebook’s new Split Test option:
Only Tests Ad Sets
Currently you can only Split Test ad sets, not creative or campaigns. That means your creative will still be self-optimized by Facebook, rather than delivered evenly amongst your audience.
No Audience Overlap
Unlike old test that showed ads to audiences within the same group, the new Facebook Split Test feature splits audiences into separate, non-overlapping groups. That means your results will be much more accurate.
Single Variable Test
Running a true A/B test requires testing only one specific variable. If you change more than one element, that’s not a true test. Facebook allows you to only test on variable for each ad set.
What Can You Test
If you have access to Split Tests, you can use three different campaign objectives:
- Website conversions
- App installs
- Lead generation
The new Split Test feature only lets you test one element. Currently, you can choose between audience, delivery optimization and placement.
Let’s look at each test.
In order to test audiences, each audience must be a saved audience – you cannot create it within your ad set. To create saved audience, go to Ads Manager, select audiences, and create a new saved audience.
You can test a minimum of two and a maximum of three audiences against each other.
You can test different placements against each other to see what combination increases results. Facebook recommends testing your custom placement combination against automatic placements, rather than creating two different custom placement combinations.
The last test you can run is Delivery Optimization.
There are more options in Delivery Optimization than within other tests. You can choose between three different elements:
- Delivery Optimization – Conversions, Impressions, Daily Unique Reach or Link Click
- Conversion Window – 1 or 7 Days
- Bid Amount – Automatic or Manual
Setting Up A Facebook Split Test
To set up a Facebook Split Test, go to Ads Manager or Power Editor and begin creating a new campaign and selecting your objective. Select the Facebook Split Test check box to start the setup process.
Select the conversion to test. We’ve chosen Delivery Optimization for our test.
Delivery Optimization settings can dramatically change advertising costs and results. For example, choosing Link clicks will deliver ads to people most likely to click, while Conversions will force Facebook to find people most likely to convert.
If you choose Audience, you’ll need to create each audience as a saved audience. You can easily do this in the audience section before creating your campaign.
If you’ve chosen Delivery Optimization or Placements as your Split Test, you’ll need to create your audience and choose your placements.
Finally, you’ll need to set a budget and schedule. This will look different than normal. Split Testing let’s you run an Even or Weighted Split.
An Even Facebook Split test divides your budget evenly amongst all ad sets.
And a Weighted Split let’s you choose the budget breakdown by ad set.
You can choose which split test you want weighted more heavily. Weighting goes from 40% to 90%.
Changes like this increase and decrease budget depending on your weighting. A 40% weighting requires a minimum budget of $1333.33 while an 80% weighting requires a budget of $4,000.
The campaign must run for a minimum of 3 days and a maximum of 14. Facebook states that this is the optimal amount of time needed to ensure statistically significant results.
The test winner is determined by the lowest cost per outcome. If you select conversion, the winner would be the one with the lowest conversion costs. Once the test is complete, you’ll be notified in Ads Manager and by email about the results.
When setting up a Facebook Split Test, you should consider the size of your audience.
Let’s say we plan to target an audience made up of 100,000 people. With three tests, that leaves 33,333 people in each audience, and a minimum budget of $1,333 that we must spend. That’s a lot of money for a small audience!
Looking at it another way, if you test optimization amongst an audience of 100,000 people, how does that help scale your campaigns? If you ran the same test amongst an audience of 1-2 million people, you could easily scale those results.
So when running a Facebook Split Test, make sure you use a large enough audience to scale the campaign.
Facebook Split Tests aren’t cheap. In order to get accurate results, Facebook requires a minimum spend amount.
For Split Tests of two, you’ll have to spend at least $800. That rises if you plan to Weight the splits. For example, a 60% and 40% split would mean a budget of $1,000 and a split of $600 and $400 respectively.
If you don’t have that amount of money, you should probably stick to the old way of split testing.
Split Tests You Should Try
Split Testing is incredibly powerful. But some tests are better than others. Here are two tests that we plan on trying.
Optimization versus link clicks
Many advertisers debate about their preferred delivery optimization method, usually between link clicks and conversions.
By testing both delivery methods against each other, you will get a better understanding of which one drives better results for you.
Choosing the right bid matters.
Automatic bids can dramatically increase ad costs. But choosing a bid that’s too small means your ads might not deliver.
With split testing, you can finally see what bid works best for your target audience. You can split the bids between a low, medium and high result to see which one performs better. That will help you decide if you want more results at a higher cost or less results at a lower cost.
So that’s how the new Facebook Split Test feature works. We hope you enjoyed it.
What do you think about the new tool? Email us at firstname.lastname@example.org to tell us your opinion.