Swarm Travel Image Multivariate Testing

Will Facebook Ads Give Us A Lower CPI?

Foursquare had $350k of free ad credits to spend on Twitter for the quarter, yet they often had trouble scaling that channel and seeing CPIs lower than $40/$50. In fact, when I came in they were hanging out at about $70/$80.

As a believer in Facebook, and the wonders of multivariate testing, I decided to see if we could unlock a better channel for our marketing spend.

I took some of the top-performing travel images (based on Twitter ad performance) and conducted a multivariate test on Facebook to see which image, CTA and message yields the most installs at the lowest cost.

Swarm multivariate 1.png

And the winner is…

swarm m ad 1.png

This guy came in at $8.59 CPI, which is way better than anything we’d seen on Twitter. You’ll notice the second from the top had a lower CPI, and enough impressions to “count” as significant. So why didn’t it win? Scale.

The winner yielded 18 installs, and this other lower CPI ad only yielded 2. So while it may be the lowest CPI, we have to consider the volume of installs. Low cost doesn’t mean anything if you’re not able to increase volume at that rate.

Conclusion: Yes, results show Facebook is a better platform for our acquisition dollars, yet we still have to focus time and energy on Twitter, since that’s where the free ad credits remain.

Recommendation: Add a small monthly budget for Facebook. Set a goal for the number of installs we want monthly from Facebook, and work backward from that, assuming benchmark CPIs of $8-10.

Erika Finnimore