3 "Must Do" Initial Tests on Facebook Advertising
Published: April 18, 2017
Author: Rebecca Soderholm
One of the most valuable assets of Facebook advertising is the ability to easily test different elements, but with all Facebook has to offer, how do you decide where to start? Here is a quick cheat sheet of impactful tests to run:
1. Creative: Lifestyle vs. Illustrated
Think of this test as starting as high level as possible, so you can determine the general direction of your imagery. Once this test is complete you can move forward with smaller (yet still important) tests like different version of CTAs, colors, text, etc.
Since Facebook cannot provide a true A/B test, you should try to replicate the set-up as much as possible. To start, ensure all elements of the ads are identical, except for the imagery. This includes landing pages, ad copy, overall concepts, and the Facebook button. Launch the two ads together in 2-3 different ad sets, so you have enough volume and variety to provide results. It is unlikely that your lifestyle and illustrated versions will have the exact same image concept, so to counter the difference in images, I recommend testing at least 2-3 different versions of the imagery to ensure your first test wasn’t a one-off winner or failure.
In my experience, it can go either way for a client (in terms of illustrated or lifestyle imagery being the most effective), so it is best practice to use tests like this to determine what works best for you!
2. Bidding: CPC vs. oCPM
This test is crucial for both performance and the ability to scale. On this test, use one audience with under 500,000 in reach and one audience with about 2 million in reach. You want two different reaches because oCPM is intended for larger audiences but has potential to perform on smaller audiences. Once you have the audiences, break them out into two identical ad sets and set one as CPC and one as oCPM. It is preferable to have these 4 ad sets running at the same time to ensure that outside factors, like an increase in competition, are not a factor in the test. Make sure that you let each test run for at least 2 weeks to ensure that the Facebook algorithm has enough time to run at highest efficiency.
3. Targeting: Top LTV Lookalike
In terms of lookalikes, advertisers tend to think it is optimal to have a huge list of customers as a seed audience. Here at 3Q we think differently. Instead, we recommend that our clients create a list of 5,000 people as a seed audience; here’s why:
- Since the audience size is smaller, it automatically forces us to narrow down the audience to the most qualified members
- If the seed list is of higher quality, that means the lookalike audience Facebook builds should contain people of higher potential
In terms of what audience to use first, I recommend creating a seed audience off of your top lifetime value customers. To determine your top LTV, you can consider different factors like amount of money spent, longevity of membership, how often they use your app, etc. Once the LTV is determined, you can then start by narrowing it down to the top 1% in increasing percentages until you acquire the 5,000 in reach. Another tip I recommend is starting off with a lookalike 1% of this audience. If it produces strong results, then open it up to a second audience, a lookalike 3% with the lookalike 1% excluded.
The number of tests you can run with Facebook ads can be a bit overwhelming, but we’ve seen tremendous impact from the results of the three tests here. If you have a favorite test, drop us a comment!