Facebook Lift Tests – Yes, They Work, and Here’s How
Published: January 22, 2016
Author: James Houchin
In theory, Facebook’s Conversion Lift Tests allow us to measure incremental lift. This is different from total conversions; basically, what lift tests set out to do is parse out which conversions you would have gotten anyway and show us what conversions Facebook was responsible for driving.
So do lift tests work? We put them, well, to the test…and we were pleased by the results.
What we tested
Our client (a B2C company in financial services) wanted to know what additional impact Facebook ads would have when run on top of their direct mail campaign. We ran the test over 8 weeks (Facebook recommends running the test for 4-6 weeks and then having a 2-week observation period for latent conversions), totaling over 6.5 million impressions.
We took the company’s direct mail audience and divided it in two; one half was shown Facebook ads, and the other was a control. We then served ads to our test group, ran the ads for the duration of the DM campaign, and measured the number of conversions from the control vs. the test group.
Why we tested
We wanted to measure the delta in conversions between an audience seeing both direct mail and Facebook ads versus those receiving direct mail only. Because the only difference between the test and control groups was the Facebook ads, we can assume that any incremental conversions were the result of our ads.
We saw a 13% lift from our Facebook ads; this means that Facebook ads drove an additional 13% of conversions. We immediately used those learnings to scale our Facebook campaigns and adjust budget with respect to our CPA goals. The data also led us to recommend running Facebook ads alongside future direct marketing programs.
The test showed that in order to make full use of the feature, we have to start with large targets (4-6 million users). Because we save half of the audience as a control, and our bid type also limited reach, the target shrunk quickly.
It’s also important to note that lift can vary by campaign. Our direct mail test was to users who were familiar with our brand (they received mail from us). A digital-only lift test may have a higher or lower lift depending on the audience.
If you or your clients are running multi-channel campaigns that include Facebook, we’d recommend setting these up immediately. The resulting data can help you make recommendations for budget allocation – not to mention highlight interplay between channels.