To be successful on Facebook, before spending one cent on ads you need a few key items in place (outside of a good product and money to spend on ads).

1. 1. Library of copy and creative to test with

   2. Strong business model with margin for scale

   3. Business metric to base success off of – IE: a target CPA

There is no point in wasting money on Facebook ads unless you have these key things in place, most businesses fail on Facebook by not having enough margin to scale or burn out quickly by not testing and having new creative to test on a constant basis.

You cannot win without testing!

For mature brands that are already scaling or scaled the same methodology applies for further scaling and testing new products and angles.

There are a number of dynamic ad products available on Facebook, however, I am specifically referring to DCO ads or Dynamic Creative ads.

DCO ads provide the best value for rapid testing on Facebook by giving the platform choices to show potential customers based on machine learning instead of forcing content that may not resonate with the target audience, this type of ad unit can be used for testing any element of an ad from Copy, Creative, Headlines and CTA’s.

When used in performance marketing these ad units can be used to test against current winners to beat them and get your current best or worst CPA down further.

This is how we do it:

• 1 Campaign – CBO – objective: Sales (We test in CBO as we are typically running multiple tests concurrently, CBO helps allocate budget to the winning adset and helps us to find winners faster).

• 1 Adset per test (You can only have one ad in each dynamic creative Adset)
• When creating the Adset, toggle dynamic creative on.
• In the audience section select only the following: Location, Gender, Age – keep this as broad as possible in terms of age and gender. (do not use any audiences, this needs to be broad)

• 1 Dynamic Ad Per Adset

When creating the ad toggle: Optimise creative for each Person, this is extremely important as when the ad builds social proof and comments FB will in some cases replace a headline with what might be a good review or a good comment that will resonate with other users.

Use a naming convention along the lines of: sku or collection name, angle, and start date. This will help identify the product, angle, and amount of time the test has been running when reviewing metrics.

We use various combinations of assets in terms of Copy, Creative, Headlines and CTA depending on what the objective of the test is.

We typically recommend at least 2 creatives, which can be video or images, We have found mixing video and images does not work as well as using the same creative type, ie all images or all video.

Note: you cannot edit placements on dynamic ads for different media types.

• 2 main text: you can test long and short copy here
• 2 headlines

So you’re going to have something like this:

• 1 Campaign set for Conversions
• 1 Adset with Dynamic Creative enabled
• 1 Dynamic ad with 2 creatives minimum, two main copy, and two headlines for a total of 6 iterations of the ad unit.

Budget-wise, if we are testing one ad we usually allocate 2X to 3X on platform CPA per day per test.

We use CBO for testing particularly when we are testing for performance, if you are only testing one ad CBO will have zero benefit, however if you are testing multiple ads simply place X amount per ad per day on the CBO – for performance testing you will find what works very quickly as the ad with most spend is likely going to win. Once you pull the winner leave the budget as is and the next winner can be found.

We usually test for a minimum of 7 days, however, the longer the ad is on test with stable results the more likely it is to scale. In general, if the dynamic ad unit is producing results at or below the target CPA continue to increase the budget until such time CPA begins to increase above the target CPA.

Note: IOS conversions can take 72 hours to show up on Facebook as Apple reserves the right to throttle conversation events, if the Apple user has opted out statistical modelling will be used by FB and this process can take the entire attribution window (IE: 7 Days) for results to show on the dashboard.

How to determine the winner?

The objective of the test is to find an ad that can deliver sales at a defined business metric, in most cases that will likely be a target CPA.

If the ad unit does not meet the business metric, the test has failed.

If, however, the ad unit as a whole does meet the metric you can easily extract the winner from all of the interactions by selecting edit at the ad level and selecting edit, share, view Facebook post with comments.

This will present a list of iterations of the ad that got the most delivery – the winning post is likely the first one you see in the list.

Note: the first post in the list is usually the most scalable, Facebook determines the winner based on interaction and social proof such as, likes, shares, and comments, however in some cases this iteration is not the one that got all the conversions – you can check this by viewing the breakdown by dynamic creative element in ads reporting.

To select the winning post, click the timestamp, the URL will change to facebook.com/xxx/posts/randomstring

The random string is the winning post ID – this can now be inserted into your scale or staging campaign depending on how your account is designed