The simplest way of avoiding to spend too much money on Facebook ads is to perform various A/B tests. And the traditional way of performing the tests can cost you a lot of money, even if you run these ads for a shorter time and on a limited audience. Now, you no longer need to spend money on many failed tests until you get the right ad set. The new Facebook Experiments tool allows you to perform these tests efficiently, at no additional cost.
Why the Facebook Experiments Tool?
Facebook acknowledges the fact that competition for ad space makes bids go higher and higher. Thus, a lot of advertisers, especially small businesses and independent marketers, are losing the race. They do not afford the bidding cost and their ads do not show, or are shown to a less qualified audience.
This would lead to a vicious cycle, where many advertisers would step away from the Facebook ecosystem (which also includes Messenger, Instagram and Whatsapp). In turn, this means reduced variety in the paid content available to Facebook users. This is a lose-lose situation for everyone, including Facebook.
Thus, using Facebook Experiments, all advertisers can perform the tests they need to determine the best performing copy and visuals for their ads using the intelligent algorithms developed by Facebook. This means that they will spend a minimum amount of money to get the answers they need and select the winning ad set.
What Can You Do With the New Testing Tool?
At the present, Facebook Experiments offers four types of tests. There are chances that, as more business accounts use it and feedback comes in, Facebook may update or change these tests.
Before you start using these tests, you need to perform a few checks of your own on the ads you want to run in experiment mode. The fist step is to make sure that the copy and visuals comply with the Facebook Advertising Policies. The tests may not indicate whether your ads are compliant, but you will note that they are flagged and do not run once you launch the campaign.
Secondly, proofread everything and make sure that the visuals are properly branded and are relevant for the message you want to promote. Now that everything is nice and tidy on your end, let’s see what kind of tests you can perform with Facebook Experiments:
1. Campaign A/B Test
You already know how this works. Now, you can do it with maximum accuracy. In Facebook Experiments you can compare several campaigns and determine the best cost per result or per conversion.
The test will split audience exposure evenly and in a random manner. Also, if you opt to measure results in cost per conversion, the test will also add a holdout.
2. Holdout Test
A holdout test is a marketing practice that allows you to test the lift of your ad campaign against a control group that does not see the ads. In this context, lift means the incremental increase in your revenues generated by the ad campaign.
Thus, a holdout test means that a group of users will not be able to see your Facebook ads in the test. Their behaviour will be compared with the group that receives the ads. If the likelihood to convert is higher for the group that saw the ads, you should run the campaign. If there are similar results with the holdout group, then this specific ad set will not generate lift.
3. Brand Survey Test
This test allows you to determine the opportunity of running brand awareness ads. It consists of two randomised groups – the group that will see the ads, and the holdout group. After running the ad tests, the audience will receive a survey with questions about your brand or products.
Thus, you will be able to determine whether your intended audience has sufficient knowledge of your brand to start running product sale ads, or whether your brand still needs to gain traction.
4. Campaign Budget Optimisation
This is a test that does not concern the content of the ad set, but how much you pay per click or impression. You need to have an existing active campaign which will be used as template in the test.
Thus, you can select various bids and Facebook Experiments will show you the impact on reach, clicks or impressions. It is very useful to perform such tests instead of interrupting a running campaign. If the test shows a negative impact to the results if you lower the bid, then it’s better to have the campaign already running then have to set it up again.