Are you still running Meta ads strategies that you used years ago? Do you ignore Meta’s best practices and recommendations because you swear that they don’t work?
My view of ad strategies isn’t absolute. There isn’t one approach that will always work for everybody in all situations. If you’ve found what works for you, great. Even if it’s inconsistent with what works most often, there are exceptions.
But, you also shouldn’t do this blindly. Don’t be stubborn about it. Don’t take an approach based on gut feel, a lack of trust in automation, or because something did or didn’t work a few years ago.
If you’ve been taking the same approach for the past year or longer, it’s important that you test your assumptions about what works and what doesn’t. And when you do, make sure it’s a scientific test that will provide meaningful results.
Running these tests can only be productive. It could reinforce what you believed to be true. Or the results may make you question whether what you’re doing is actually effective. You may see an alternative approach in a new light.
My advertising approach has changed dramatically over the years. I did not immediately embrace an evolving set of best practices. I was stubborn. But, my own tests have helped me understand that I was wrong. They also helped improve my confidence in another approach.
In this post, we’ll cover a handful of old school advertising targeting strategies and how you should test them against a more modern approach. Once you’ve tested, you can decide whether your stubbornness was right all along.
Testing Basics
Before we get to the old school strategies, it’s important to provide a framework for testing.
1. Use A/B Test.
I prefer to create A/B Tests in Experiments. Create the campaigns or ad sets that you want to compare first. Then go to Experiments and click to create an A/B Test.
Select the campaigns or ad sets that you want to compare. I ask you to test ad sets in two of the three examples below. In the third, you’d compare campaigns.
2. Focus on a Single Variable.
Everything about the two campaigns or ad sets should be identical except for a single variable. Since this post is about testing targeting strategies, everything beyond targeting should be the same. Make sure that there aren’t any other variables like placements or ad copy and creative that could result in differences in performance.
3. Your Key Metric
The Key Metric is what determines which campaign or ad set “wins” in an A/B test.
Make sure that this metric isn’t frivolous. What ultimately determines which ad set was better? If your goal is sales, then the key metric should be Cost Per Purchase. Do not use secondary metrics like CTR or CPC.
If your key metric is Cost Per Lead, you may want to take steps to measure the quality of those leads. Make sure that you send these leads to different forms so that you can keep track of them in your CRM.
4. Strive for meaningful results.
Your goal isn’t to find a winner quickly, it’s to find convincing results that actually mean something. Make sure that the budget dedicated to each competing campaign or ad set, combined with the length of the test, are enough to produce the volume that you need.
The longest you can run a test is a month. This would be my preference for a test that will help define your strategy going forward. Do not end the test early if a winner is found.
If the results become more convincing with time, that’s a good thing.
1. Interests and Lookalikes
There was a time when the ability to target people by interest, behavior, or lookalike audience was revolutionary. It gave advertisers targeting control and your ads were more likely to reach a relevant audience.
That isn’t always the case now. If you use Advantage+ Audience, any inputs you provide for detailed targeting or lookalike audiences will be suggestions.
This is why many advertisers have resorted to using original audiences. Targeting inputs in that case are more than suggestions — or we assume.
But, the reality is that even when using original audiences, your targeting inputs are rarely tight constraints. If you’re optimizing for conversions, link clicks, or landing page views, Advantage Detailed Targeting is automatically on.
If you optimize for conversions, Advantage Lookalike is automatically on.
In other words, we have no idea how much your selection of those interests and lookalike audiences actually matter. And based on my tests, they matter very little — if at all.
It’s not even clear that your audience suggestions matter when using Advantage+ Audience. They may actually be a detriment. This is why I recommend testing your current strategy with interests and lookalike audiences versus Advantage+ Audience without any suggestions at all.
Compare:
- Version 1: Original Audiences using Detailed Targeting or Lookalike Audiences
- Version 2: Advantage+ Audience without Suggestions
Key Metric: Cost Per Conversion (whichever event is most relevant)
Are you actually better off using original audiences to target interests or lookalikes? Maybe. But, prove it.
2. Gender and Age Control
One of the complaints I’ve heard from advertisers about Advantage+ Audience is the lack of control over age and gender.
You are only able to provide an age minimum within Audience Controls when using Advantage+ Audience.
Any age maximum or gender inputs you provide are audience suggestions. If Meta can get you more or better results by delivering your ads outside of those ranges, it will.
As a result, advertisers who feel these inputs are critical have favored original audiences. In that case, age and gender are tight constraints that will be respected.
Let’s assume that your customer is predominantly women aged 25-49. If Advantage+ Audience works the way that it should, whether or not ads are delivered to men or people outside of those age ranges will depend upon whether you can get your optimized actions from those other groups.
I’ve seen examples where businesses that serve women used Advantage+ Audience and 99% of the budget was spent on reaching women — even though gender is only a suggestion.
The key, though, is that you should optimize for conversions for this to be effective — preferably purchases. If reaching people who fall outside of expected gender and age range won’t lead to conversions, you’ll likely spend very little there.
Can you trust Advantage+ Audience without these controls? It’s worth testing for any type of conversion, especially purchases. Leads can be problematic since it’s possible you may get cheaper and lower quality leads this way — but, it’s worth testing. Engagement optimization is likely to go off the rails using Advantage+ Audience without those controls, but top-of-the-funnel optimization is problematic at its core.
Compare:
- Version 1: Original Audiences with Age and Gender Restrictions
- Version 2: Advantage+ Audience with Age and Gender Suggestions (if at all)
Key Metric: Cost Per Conversion (whichever event is most relevant)
Is it critical that you only reach people within your preferred demographic? Is it possible that Advantage+ Audience will waste money by reaching people outside of those groups? Maybe. But, prove it.
3. Remarketing
Look, my whole thing years ago was remarketing. I was generating a high volume of daily organic traffic, and ads allowed me to leverage this with highly relevant targeting.
But, things have changed. You can still target remarketing audiences. Those groups of people are surely just as relevant as they were years ago. What changed is the cost.
Targeting small groups of people is much more expensive than targeting large groups. Even though your website visitors may be three times more likely to convert, it may cost three (or five or 10) times more to reach them.
The other development is that Meta’s ad delivery algorithm has improved. Even if you use Advantage+ Audience without suggestions or go broad with original audiences, the algorithm will almost always prioritize a percentage of your budget to remarketing. We now know this due to Audience Segments.
When running Advantage+ Shopping Campaigns (or any Sales campaigns, if you have the update), you can breakdown your results by Audience Segments. I’ve often seen that between 25 and 35% of my budget is spent on people who have engaged with me (visited my website or subscribed to my email list) or bought from me.
Many advertisers continue to create campaigns with separate ad sets for prospecting and remarketing. But, since these two things happen at once without us even realizing it, is it still necessary?
For this test, we’ll need to compare campaigns since the old school approach is to use two ad sets. I would also use an attribution setting that is click only to prevent the remarketing ad set from inflating results with view-through conversions.
Also make sure that the combined budget of each campaign is the same. In other words, Version 2 using Advantage+ Audience should be the same as the sum of the two ad sets in Version 1.
Compare:
- Version 1: Campaign with Remarketing and Prospecting Ad Sets
- Version 2: Campaign with one Ad Set using Advantage+ Audience
Key Metric: Cost Per Conversion (whichever event is most relevant)
In addition to comparing the Cost Per Conversion, use your Breakdown by Audience Segments to see how your spend and results from remarketing compare.
Test Your Assumptions
I want you to test these because what I’ve seen from my own tests is quite clear. I’ve seen that…
1. Detailed targeting and lookalike audiences are rarely beneficial. Advantage+ Audience almost always gives me the same or better results.
2. Gender and age restrictions are rarely necessary. Especially when optimizing for purchases, the algorithm figures it out.
3. Remarketing is not the advantage it once was. It’s expensive to run stand-alone remarketing ad sets. Remarketing and prospecting happen together in the most optimal way now.
There are always exceptions, and I’ve even mentioned some of those cases in this post. But, if you are still utilizing some of these old school targeting strategies, I encourage you to run these tests yourself and allow for the possibility that more modern approaches may be more beneficial.
Your Turn
These are the types of tests that I often run to challenge my own assumptions. Once you’ve run these tests, I’d love to see your results.
Let me know in the comments below!
