11. Split Testing

When you launch a Facebook ads campaign, do you split test? Do you create multiple ads with variations based on targeting, bidding, copy, imagery and ad type to determine which works best?

You should!

Ready? Let’s go!

What You Are Going to Learn Today

Today’s lesson is Split Testing. We’re going to focus on the following:

  • What Split Testing Is
  • How Split Testing is Changing
  • How You Should Split Test

The Lesson

What Split Testing Is
Split testing is the process of creating multiple ads or campaigns and testing the results in an effort to find which combination of copy, imagery, targeting and bidding are most efficient.

Since there is no magic pill — no perfect combination that will always work — it’s important to try out various combinations.

And if you only create one ad and rely on that one ad’s success, your opportunity for failure is huge.

How Split Testing is Changing
For the longest time, I’ve created campaigns with many different ads in them. But the necessity to split test for many of these items no longer exists.

The reason for that is the new Facebook ad reports. We’ll be digging into these reports more in the bonus section, but because of these reports you can find out some pretty impressive information from a broadly targeted ad.

For example, you can create one ad that is set up as follows:

  • All Placements
  • Multiple Countries
  • Male and Female
  • All Ages

In the past, I may create ad variations for each of the placement options. Or for individual countries. Or one for male and one for female. Or one for several age segments (18-24, 25-34, 35-44, 45-64, 65+).

But with the new ad reports, Facebook can tell you cost per desired action (and a whole lot more) broken down by placement (sidebar, desktop news feed, mobile), country, age group and gender. All from a single, broadly targeted ad.

Of course, that doesn’t completely remove the need to split test. But you no longer need to split test for those four things.

How You Should Split Test
Different advertisers will have different methods of split testing. But I’d like to share mine that you can rework for your own needs.

I like to create individual campaigns that have ads in them that are all very tightly related. They need to have the same goal and be the same ad type, but with minor variations.

First, following are the types of things I continue to split test:

  • Copy (if applicable)
  • Imagery (if applicable)
  • Targeting segments (Precise Interests, Partner Categories, Broad Categories, Education, Workplaces)
  • Connections (Fans, Friends of Fans, no connection)
  • Custom Audience Segments
  • Lookalike Audience Segments
  • Bidding (Optimized CPM, CPM, CPC)
  • Conversion Specs (optimized for link click, engagement, photo view, etc.)

My recommendation would be to create separate campaigns for each of these segments. You may have three different copy variations, three different image variations, etc.

Of course, this is a lot of work. I don’t split test for all of these things on each promotion. But these are the things I consider when split testing.

When putting together your ad groups, much of what you’ve already learned comes in handy. Make sure you use the “Duplicate” button to duplicate similar ads and make minor changes. And use the Saved Target Groups you created last lesson to easily split test your target segments.

Your Assignment

Action Items:

  1. Create a campaign promoting a post based on bid options
  2. Create a campaign promoting a post based on targeting segments
  3. Create a campaign promoting a domain based on copy and imagery options
Need personal one-on-one help from Jon?
Set up a training session!