Skip to main content

So, you want to run some ads on Facebook. But how do you determine if the ads you run actually achieve your goals?

Quick Answer: If you see more results based on the objective you set for a campaign, you may already be doing a good job, but if you want to be sure, test.

Facebook Ad Testing

Enter Facebook ad testing—a great way to ensure that your ads hit the mark based on your objectives.

There are two general types of tests you can run on your Facebook ads to make sure you get the results you want: A/B Testing or Lift Testing. Once you finish the tests, you can use the results to see where you can make improvements to the ad campaign.

Facebook A/B Testing

How Facebook Runs A/B Tests

On Facebook, this means that your ad audience is divided into random, non-overlapping groups to reduce that impact from outside factors on your ads.

What can you test with A/B testing?

You can run similar variations of an ad to see which works better. For instance, you can switch up the ad’s imagery, use the same copy, and use the same call to action (CTA). Then, review the results to see which ad “won.”

It is crucial to keep in mind here that you should only change ONE variable for testing. Remember way back when you changed one thing in high school chemistry, and well, sometimes fire happened? Yes, we’re looking for that spark in your ads. Run the ads for a short period of time (a week is good) and see what results shake out.

How Can You Tell Which Ad Won?

The winning ad set is determined by the Cost Per Result (CPR) based on your campaign objective. If your objective is link clicks to the landing page you have specified, you would look to see which ad had a lower cost per result regarding clicks or cost-per-click (CPC).

Facebook Lift Testing

How Facebook Runs Lift Tests

On Facebook, a lift test compares the actions of people who have seen your ads vs. those who have not to determine if ads generate the desired outcome. A lift test should be used when your goal is to measure brand recognition and conversion from brand recognition. The more “lift,” the more effective your ads are.

There are two types of lift:

1) Brand lift: Brand lift measures and assesses incremental brand awareness through polls and questionnaires. If you have ever come across a simple survey in your Facebook news feed that asks if you remember seeing a specific ad for a particular brand in your news feed, chances are you were selected to be a part of a brand’s lift test (lucky you)!

2) Conversion lift: Conversion lift is defined as incremental conversions assessed through user actions gathered from a website or app using standard and custom events. It compares additional business (online, offline, or on an app) driven by advertising on Facebook across Facebook, Instagram, and Facebook’s Audience Network across devices. By making this comparison, you can get a bigger picture look at ad performance (*Note: You must install a Facebook Pixel on your site or have the SDK in place for in-app events.) “Events” are actions that happen on your website or app. They can be actions such as adding payment information at checkout, adding an item to a cart, completing an event registration, scheduling an appointment, etc. Facebook provides a list of predefined, standard events that you can choose from.

An example of conversion lift would be when someone sees your ad on Instagram but does not click or take any action to complete a purchase on your website. Then, that same person ends up visiting a website that is a part of Facebook’s Audience Network. Now, that same person sees the ad again and actually clicks on the ad. That same person is now directed to your landing page and, voila, they make a purchase. To see if there is lift attributed to seeing the ad across the Facebook Audience Network, we look at the difference in people just making a purchase versus people who see the ad and then make a purchase.

Say, during a typical business day, on average, only five customers visit your online store, and two complete a purchase. However, after running a conversion lift test with an ad on Facebook, you see that there are, on average, ten customers visiting the store, and now seven are making a purchase. Facebook would then attribute the conversions to the ads because they are tracked across devices. Facebook would calculate lift and provide a test confidence percentage to deem the test statistically reliable.

What is Facebook’s Confidence Percentage?

“Confidence percentage” is a term used to describe the likelihood that your ad would receive the same results if the test were to run again. A strong confidence percentage that meets or exceeds a specific benchmark deems your test statistically reliable. According to the confidence percentage, a statistically reliable test can then be said to provide the same results in a real-life scenario or if the test is rerun. If the confidence percentage is less than what is deemed successful, you would not want to run this test again and instead you would make changes to make the test more reliable and then re-run the test. Note that the more data you collect or the more people you have interacting with your ad will help increase the probability that your test results will be statistically reliable.

For an A/B test to be successful, the confidence percentage should be at least 65%. Learn more about A/B Testing on Facebook here.

For conversion lift campaigns to be deemed successful, they should receive a confidence percentage of at least 90%.

As you can imagine, running tests on Facebook is essential to making sure you are using your budget wisely to create impactful ads that get results. You can try it on your own. Or, make it easy on yourself and partner with the CloudControlMedia experts to simplify and streamline your Facebook strategy.

Lindsy Bentz