Facebook split tests let you hone in on what’s working and what isn’t for your Facebook Ad campaign. Want to improve your ROI? Split tests are the way to go – because how else are you going to prove your hypothesis correct.
You may have spent a good time carefully considering your targeting, but how do you know you’re reaching the right people? Would you get better results if you narrowed your audience? What about the ads themselves – are you getting the most out of them?
This is where you’ll want split test different aspects of your campaign. Usually, these split tests fall into the following categories:
- Ad types
- Campaign objective
But what should you test? In this post, we’ve listed a selection of elements you can split test. Depending on your budget, you’ll want to prioritise which split tests to run first.
1. A/B test your saved audiences
Facebook’s Audience Insights is a neat tool that allows you to create different audiences and save them later for your campaigns. Here you can try out different targeting options and identify what makes sense to test.
For instance, in the screenshot below, I’m setting up a campaign to target accountants in the United Kingdom. When it comes to demographics, women and men aged 25-44 are the ones most likely to engage. The blue bar represents how your selected audience would perform when benchmarked to Facebook’s averages (grey bar) – so whenever the blue bar is above the grey, this means that group has a high chance of clicking on your ad.
In this case, it might be a good idea to split test based on age and gender.
There are other audience features you can look at and test too, such as:
- Age and gender
- Relationship status
- Top countries
- Education level
- Job role
- Top categories
- Page likes
- Top cities
- Frequency of activities
- Device users
- Top languages
2. A/B test Custom vs Lookalike
First, let’s clarify what a Custom Audience and Lookalike Audience are.
A Custom Audience is an ad targeting option that lets you reach people who already know you. A Custom Audience is usually created by uploading a list of your existing customers. Alternatively, it can be based on people who have visited your website, which is tracked through the Facebook Pixel
A Lookalike Audience is based on your Custom Audience or on the people who have already converted during your campaign. The aim here is to reach new people who are most likely to convert because they share similar traits to your existing customers.
As a split test, you can use these two audiences against each other to see which one performs better.
3. A/B test your Facebook ad types
In total Facebook has eleven different ad types for you to use. Those include:
- Instant experience
- Lead generation ads
- Post engagement
- Event responses
- Page likes
That’s a nice selection of ad types to choose from, which makes it perfect for a split test.
Depending on your campaign objectives, you will be to find at least two ad types you can use.
Let’s say you’re organising an event the most likely ad type you would go for would be “Event response”, but you may also want to try video and image ad types too. This way you’ll find out which ad type performs best for you.
4. Split test images vs videos
I get it. Videos aren’t easy to produce unless your professional at it. But should that stop you from testing between images vs videos? No, it definitely should not!
What’s more is that according to the Aberdeen Group, using videos can help you achieve a 27% higher CTR and 34% higher website conversion rates.
So before you dismiss videos, ask around to see there’s any material you can use. If you can find something to edit, then great. If not, you may want to turn to a tool called Lumen5.
Lumen5 enables you to transform your content into videos in minutes with the help of artificial intelligence. In essence, it’s video creation on auto-pilot.
5. A/B test long vs short copy
When it comes to ad copy, some marketing professionals say the shorter, the better.
But how do you know it beats longer ad copy? Sometimes, the longer ad copy will outperform the shorter text. You won’t know until you test it. Here you can see an example from CRM software provider Pipedrive testing this assumption out.
Essentially, you’ll want to have your pitch nailed down and try different variations of it to see what sticks.
A handy resource to check out is this blog post from Valuer: Why Your Elevator Pitch Sucks…
We’ll be adding more Facebook split tests and updating this post regularly, so stay tuned by liking our Facebook Page