When we avoided these three Facebook Ad mistakes, ROAS improved by 4x

With this client, we worked in parallel to improve the three factors highlighted at the beginning;  Campaign Structure, Creative, and Targeting.  

If you are a marketer struggling to improve the Facebook Ads performance, I recommend reviewing your Facebook Ad account immediately after reading the blog.

Top 3 Mistakes Marketers Make with Facebook Targeting Settings

Targeting is a major variable and a contributor to Facebook Ads performance. That’s why you need to make sure that you have it exactly right.

If the settings remain suboptimal, it could be a major factor that hinders your Facebook Ads performance. Your Targeting settings is an area that requires attention.  

Let’s dive in the top 3 mistakes we’ve identified that could potentially generate a significant boost to your campaign if fixed.

Mistake #1 Over segmenting your audience

Targeting is set up on a per ad set basis. We have seen cases where the ad sets are based on granular categorized targeting.

The most common ones are:

• Retargeting on recency is subdivide into 1 day, 7 days, 15 days, etc.
• Subdivide the audience age into five-year increments.

Many marketers feel the retargeting should be broken down into granular segments when adjusting the bids. Now, more than before, the bidding logic is very different.

The major difference in the current bidding strategy from before is much of the bidding strategy is controlled by machine learning.

For many cases, excessive segmentation is incompatible with machine learning.

What’s important is targeting that fosters smart machine learning.

A key feature of a machine learning bidding strategy is that machine learning learns performance data and automatically bids on target.

There is one thing you absolutely must keep in mind when bidding with machine learning.

That is; You have to let it learn with enough data for it to make optimal bidding decisions.

Machine learning learns on a per ad set basis. If the ad set is overly fragmented, it isn’t easy to accumulate enough training data. As a result, optimization will not work well.

A similar situation applies to humans. If you want to improve your grades, you study a lot. The same is true for machine learning. To get smarter, you have to study and learn a lot.

The goal is 50 conversions!

Facebook’s guidelines recommend getting “50 conversions per week per ad set”. Anything less than this, your ad performance will be inconsistent due to insufficient training data.

Aim for “50 conversions per ad set per week” and try to consolidate ad sets targeted at similar audiences into one.

Many marketers mistakenly over segment their audiences with respect to things like device, ad placement, platform, gender, age and more.

This seems like it may be a good idea since you’ll have more granular control over each audience segment and an easy way to quickly see various performance metrics for each audience.

The truth is, you’ll actually be heavily inhibiting Facebook’s machine learning algorithms to effectively optimize your campaigns. In other words, you’ll be missing out on more money for your business.

It’s important to strike a balance between the trade off of granular segmenting and feeding machine learning algorithms proper conversion data. It’s recommended to feed a minimum of 50 conversions per week per ad set so that Facebook’s algorithm can properly learn.

Remember that many of the ways you might set up ad sets to measure audience segments in detail can actually still be measured by using Facebook’s dimension breakdowns reporting feature. This feature will allow you to see data for your campaigns by any of the following dimensions:

• Age
• Gender
• Country / Region
• Device
• Ad Placement
• Platform
• And many more…

You might be tempted to separate ad sets by some of the dimensions listed above but we recommend avoiding this unless absolutely necessary. If you’re in doubt just set up an experiment to collect actual data showing which set up performs better.

We recommend that instead of separating the ad sets by any of the dimensions above, just use the dimensions breakdown feature to see how your ads are performing in terms of each dimension.

If you’re feeding the algorithm enough data, then you’ll see your spend automatically being allocated towards the segments in each dimension that perform best.

Occasionally the algorithm may get things incorrect and so it’s still important to monitor your performance across these dimensions and intervene only if necessary.