Why did this happen?
When I used Advantage+ Audience without audience suggestions, it resulted in Meta spending about 35% of my budget on remarketing to my audience segments.
When I provided audience suggestions using custom audiences that matched my audience segments, it resulted in 32% of my budget spent on those groups.
I was determined to figure out what this means — if anything.
Taken Separately: Encouraging
I was initially stumped by these results. Separately, these results were both encouraging.
The fact that Meta spent 35% of my budget on remarketing is an incredible discovery. It vastly improves my faith in taking that approach. I understand why advertisers may not trust a hands-off approach, but this is evidence that it does exactly what we want.
And it was also good to see that Meta spent 32% to reach the audience I provided as a suggestion before going broader. That’s how Meta says it should work, after all.
The Problem
While this is all great, the fact that more of those people were reached without suggestions gave me pause.
Is it possible that these suggestions were mostly ignored and they would have been reached anyway? Did the fact that I provided suggestions actually negatively impact my ability to reach those people? It’s only a 3% difference, but it’s a difference nonetheless.
Other Findings
First, CPM was five dollars lower for prospecting when I didn’t provide suggestions. That could be random or by design. By not providing suggestions, it may give the algorithm more freedom, which could lead to lower CPM costs.
Second, I should note that the percentage of impressions dedicated to remarketing actually flipped the other way.
23.7% of impressions went to my audience segments when I didn’t provide suggestions.
29% when I did.
That’s also a bit of fun with numbers. The reason that flip happened is at least partially because CPM was higher to reach the prospecting audience when providing suggestions.
Regardless, the suggestions made very little difference, and that’s the most important takeaway. Why?
My Thoughts
The fact that suggestions didn’t seem to make any positive impact is likely because I have an established ad account with pixel, conversion, and ad engagement history. Since I’m optimizing for a conversion event, Meta has plenty of history to go off of when assembling an initial target audience for me. Suggestions aren’t necessary in that case (or so it seems).
On the other hand, suggestions may be most useful when that history doesn’t exist. New ad accounts, new pixels, or websites that generate minimal activity are likely at a disadvantage. Meta may need some help getting started.
Obviously, the results from my test remain a drop in the bucket. You never want to look too far into small sample sizes. It’s always possible I’ll see something different in future campaigns when I do or don’t provide suggestions.
But these initial findings are certainly interesting and something to watch.
