My Experience with Facebook's Targeting Expansion

(Originally posted 24 September 2019 on LinkedIn)

Facebook’s targeting expansion is certainly not a new function, but as I am executing paid social for a boutique fitness studio, I thought I’d pen my thoughts and findings as they happen to document my progress over the months.

About Targeting Expansion

According to Facebook, 

Targeting expansion helps improve your campaign performance by allowing our system to reach a broader set of people than those you defined in the detailed targeting section. 

You can use this option when you want us to show your ad to additional people who we think would get you more and/or cheaper results.

The option of targeting expansion appears under the ‘Audience’ portion of an ad set:

Experiment Setup

At the start of this month, I ran a campaign which drove traffic to different pages of the website. It was optimised for landing page views, and I created a lookalike audience from a seed audience of top spenders.

By mid-September, I got a notification asking if I would like to try out targeting expansion in a split test. I decided to do so, and Facebook duplicated an ad set (within the same campaign) with the only variable as allowing for targeting expansion.

Results

The test started on 18th and results were in by the 22nd. And yes, Facebook concluded that the one with targeting expansion returned with a lower average cost per result ($0.85 vs $1.13)! 

But I had a hunch that the cost per result for the control (‘Without target expansion’) drastically increased during the course of the split test, and so I proceeded to review its performance before the test began on the 18th. 

True enough, the cost per result was lower during this period of time before the split test began. This increase from $0.73 to $1.13 is a whopping 55% jump. 

Aside from this, the CPC is another mystery altogether. Although the split test has increased the average cost per result, it has actually yielded lower CPC ($0.21 & $0.19 vs $0.32). 

Final thoughts

Of course, this split test was not conducted perfectly -- for one, it should have been conducted from the onset. Also, while targeting expansion aims to lower costs, there is the question of whether an expanded audience will ultimately bring value to one's business.

But I would love to hear your thoughts on this little experiment, how I might find some answers and what I can do to improve paid social executions in the future!