We’re entering a new era of ads optimization driven by broad targeting and automation. While this has long-term potential for sales campaigns, there are obvious weaknesses outside of that focus. How does Meta improve optimization to work in all situations?
We also want a solution that is in line with Meta’s current philosophy and campaign creation flow. We can’t add a bunch of manual inputs to improve results when Meta’s entire focus is on streamlining the process. We want to find a realistic solution.
Let’s first review the problem and an example of ads optimization gone wrong. Then I’ll provide a recommended solution…
The Problem
The flaws and weaknesses of Meta ads optimization can be traced to one simple point: The goals of optimization are literal.
What I mean is that Meta’s systems have tunnel vision. When you set your performance goal and (if necessary) conversion event, the algorithm’s primary focus is getting you that thing. This is how you’ve defined success for yourself, so Meta will deliver your ads and make adjustments to satisfy it.
This works great at the very bottom of the funnel. Not only can you optimize for number of purchase events, but you can optimize for the value of those purchases. Meta’s focus will be on satisfying that, which puts the algorithm and Meta on the same page.
This can also work for conversion leads optimization. Meta’s focused on generating quality leads who end up buying from you. That’s what you want, too.
It all goes downhill from there.
If you optimize for link clicks, landing page views, ThruPlays, post engagement, Page likes, or even leads, the advertiser and Meta’s ad optimization will have divergent goals.
Yes, the advertiser wants that initial action. But they want that initial action because they have another ultimate goal. And they’re usually optimizing for that initial action because they don’t have the budget to optimize for that ultimate goal.
But Meta’s optimization doesn’t see it that way. It’s only trying to get you as many of that one action as possible within your budget. There is no concern for quality or what these people will do later.
That’s a problem for ads optimization generally. But it becomes a bigger problem in this world of broad targeting.
Example of Optimization Gone Wrong
This blog post was inspired by a complaint I’ve been hearing a lot lately related to Advantage+ Audience. The situation that keeps coming up is advertisers promoting a brand or product that’s catered to women.
They create a campaign optimized for Post Engagement.
Within Advantage+ Audience, the advertiser defines their target audience by detailed targeting and gender.
Of course, these are only targeting suggestions since we’re using Advantage+ Audience. The algorithm can go beyond these suggestions.
So, just select a gender constraint in Audience Controls, right?
Nope, gender isn’t an option. If Meta believes that more engagement can be found by reaching men, it will show your ads to men.
This is intentional. Meta’s documentation on Advantage+ Audience only mentions the ability to exclude ages or locations.
You can imagine how this could lead to disastrous results for a brand focused on women. The inability to exclude men should be fine when optimizing for purchases. If men don’t purchase, the algorithm learns and doesn’t show to men. But you can imagine that men will engage with ads featuring women. And unfortunately, it will be some of the creepiest engagement.
Meta’s ads optimization doesn’t care whether it’s creepy engagement. It only cares that there’s engagement. And that means that this ad will be shown to more men.
The Solution: Ranking Actions
We need Meta’s optimization goals to be in line with our long-term advertising goals. How do we fix this?
The easy solution to the problem above may be to allow the exclusion of men in Audience Controls. But that’s only a Band-Aid that doesn’t solve the underlying problem.
Meta’s optimization needs to have a foundational understanding of what we want. Yes, we want engagement. But we want engagement from relevant people who could potentially become paying customers.
There needs to be a reworking of the algorithm. To do that, we can rank what’s most important to us.
For example, our performance goal may be Post Engagement, but our ultimate goal is a purchase. And if not a purchase, a lead. So, Meta’s optimization will prioritize engagement that eventually results in a purchase or lead.
How do we do that? Well, having the advertiser rank their priority events in the ad set would never fly. This conflicts with Meta’s desire to simplify campaign creation.
Maybe this could be an addition to Ad Account Settings. We’ve already seen this for account-wide exclusions related to Advantage+ Shopping and manual sales campaigns (Customer Acquisition). It wouldn’t be crazy to allow advertisers to prioritize events there.
But, this is also something that is mostly universal. Meta’s ads algorithm should be smart enough to prioritize this for us. This ranking should apply for most advertisers:
- High-Value Purchase
- Any Purchase
- Conversion Lead
- Any Lead or Registration
- Deep website engagement (time spent, return visits, events fired)
- Deep page engagement (long-time follower, quality DMs that aren’t reported or ignored)
- Deep post engagement (watch videos to completion, share posts, prioritized reactions like Love, quality comments that aren’t marked spam)
- All other light-touch engagement (clicks, reactions, views)
To clarify, this will act as a foundational ranking for helping the algorithm learn. Even if your performance goal is Post Engagement, it will focus on satisfying that goal. But the optimization will prioritize engagement that comes from people who have or will eventually perform these other actions (greater weight added based on order).
Not all post engagement is created equal. That’s obvious. We just need Meta’s ads optimization to understand that, too.
The Only Other Direction
This is something that must be fixed. Yes, optimizing for top-of-the-funnel has always been a questionable strategy, but it becomes completely worthless when the push is to go broad. Unless Meta’s optimization gets smarter about what quality engagement looks like, there’s only one other option.
If Meta can’t fix it, then eliminate it. Advertisers are burning money, and many don’t realize they’re doing it. The only other direction that makes sense is to only allow for conversion optimization. Because this is the only time when the new brand of Meta advertising makes sense.
Your Turn
Maybe this is a pipe dream, but these are critical issues that can be addressed and it would drastically improve advertising results. What do you think?
Let me know in the comments below!