3 Holes in Existing Customers Exclusion for Meta Ads

Advertisers often complain about paying to reach people they believe should have been explicitly excluded using custom audiences. The assumption is that Meta has chosen to ignore exclusions. But, the effectiveness of these exclusions is mostly within our control.

We most often hear this related to existing customers. There are two primary scenarios where this comes into play:

1. Advantage+ Shopping using an Existing Customer Budget Cap.

Advantage+ Shopping Campaigns allow you to set a cap on how much you will spend on existing customers.

Existing Customer Budget Cap

This approach relies on the definition of your Existing Customers audience segment within your Ad Account Settings.

Existing Customers

2. Manual campaign with a custom audience exclusion.

You can also run a manual campaign and exclude your existing customers by listing out the custom audiences that reflect that group.

If you use Advantage+ Audience, that would be within the Audience Controls.

Audience Controls

If using original audiences, you can exclude custom audiences.

Exclude Custom Audiences

But, even if you use these settings, you will probably still reach some of your custom audiences. Why?

Here are the three most likely reasons (along with a myth about audience expansion)…

1. Completeness and Accuracy of Data Provided

In order to exclude every existing customer, you must first completely and accurately define your customers with custom audiences so that Meta can do just that. But, this is far more difficult than it sounds, approaching unreasonable.

Here’s an example of how I’ve defined my existing customers…

Existing Customers Audience Segment

It’s a mixture of data file custom audiences and website custom audiences. But, I guarantee it’s incomplete.

To troubleshoot, ask yourself these questions…

Do your excluded custom audiences actually include existing customers?

It may seem like a silly question, but one of the first mistakes that advertisers make in this area is that they mess up the parameters that define a group of people. Look no further than inflated conversion reporting happening because the Purchase event is firing for the wrong stage.

Do your excluded custom audiences exclude all customers or only some?

When creating a custom audience based on your email list, have you confirmed that you’ve included every customer for every product? All customers historically, or only a specified period of time?

I should also point out that, depending on how you interpret Meta’s Custom Audience Terms of Service, you may be required to remove customers who have opted out of your list. So, there may be paying customers who you can’t include in the custom audience.

Custom Audience Terms

This may be pointing out the obvious, but website custom audiences are capped at 180 days. If you exclude your existing customers using this approach and your business is more than six months old, the audience will be incomplete.

Website Custom Audience Purchase

And of course, there’s a long list of potential issues with website custom audiences and completeness. The most obvious is iOS opt-outs. Meta specifically said that the result of opt-outs would be smaller custom audiences.

iOS 14 Opt-outs Targeting

That will create holes in your exclusions.

2. Meta’s Ability to Match the Audience

This mostly applies to data file custom audiences, where you provided a customer list to create a custom audience. Just because you uploaded a customer list that includes a specific person doesn’t mean that Meta will be able to match that customer’s details to a Facebook profile.

Match Rate

If you only include a list of email addresses, they need to be matched to Facebook users who provided those same addresses in their profiles. Facebook profiles may be old and outdated. Maybe your customer used a business email address that isn’t associated with their profile.

The more columns of data you provide for first name, last name, email address, phone number, and physical address, the higher the match rate will be. But, you can guarantee you won’t get a 100% match rate.

Facebook Custom Audience Data Email

It’s anecdotal, but advertisers tend to see anywhere from 20 to 70% match rates from customer lists. The ability to match is only as good as the completeness and accuracy of the data. But even then, it’s not guaranteed to match a Facebook profile that’s used for exclusions.

You could also make the argument to include website custom audiences here. If a user is blocking cookies, browsing incognito, or using other privacy settings that impact the data that can be sent back to Meta (not to mention iOS opt-outs), Meta’s ability to match and exclude users is impeded.

3. Meta’s Ability to Actually Exclude Them

This is more theory than reality, and it assumes that the source of the problem isn’t #1 or #2 above. Essentially, it would mean that despite accurately and thoroughly defining your existing customer custom audiences, you are still paying to reach the people you shouldn’t. Meta knows that a specific person falls within your exclusions, but you reach them anyway.

Maybe it’s due to a bug. Maybe it’s because Meta doesn’t care about your stinking exclusions.

I’m not saying that this is impossible. But, of the three possible explanations, it’s the least likely. It’s also very difficult, if not impossible, to prove.

By “least likely,” I don’t mean that bugs rarely happen or that Meta is always trustworthy. I mean that there are so many obvious reasons for holes in exclusions, we don’t really need to resort to conspiracy theories to explain them.

The Expansion Myth

I’ve seen the theory that audience expansion doesn’t respect your custom audience exclusions. Specifically, this is related to using original audiences when Advantage Detailed Targeting or Advantage Lookalike are turned on.

The way I understand it, the source of the theory is this passage in Meta’s documentation related to Advantage Detailed Targeting

Advantage Detailed Targeting

And a similar passage from Meta’s documentation related to Advantage Lookalike

Advantage Lookalike Exclusions

For Advantage Detailed Targeting, Meta says that you can still exclude “targeting selections outside of detailed targeting (such as age, gender, location and language).” For Advantage Lookalike, “you can add targeting selections as exclusions if you don’t want our system to consider certain demographics such as Locations, Age, Gender etc.” Meta didn’t mention custom audiences!

But, is this an intentional omission? In both cases, it’s clear that Meta isn’t providing an exhaustive list. “Such as” language when listing out what can be excluded from Advantage Detailed Targeting and an important “etc.” to wrap up exclusions for Advantage Lookalike could suggest, maybe, that custom audience exclusions aren’t respected.

I’m not buying this argument. You can still exclude custom audiences in either case. It’s far from definitive that the reason you can still reach some of these people is due to expansion.

According to this theory, the proof is that if you optimize for a top of funnel action that doesn’t require expansion, third-party reporting tools show that you reach fewer existing customers as a result. But, this is less a function of the incredibly low quality results you get from top of funnel optimization than any proof that the exclusion works in this case.

If you’re still not convinced, look no further than Advantage+ Audience. Audience Controls are where you set the specific parameters that Meta will respect. These are not suggestions, but tight constraints.

One of those controls is excluded custom audiences.

Advantage+ Audience Audience Controls

If you believe that your custom audience exclusions aren’t respected when using original audiences when expansion is on, then maybe you should use Advantage+ Audience instead. This seems backwards, though, since the entire benefit of Advantage+ Audience is that the algorithm has more freedom to reach people who are likely to convert than when using original audiences. It would be odd if it were Advantage+ Audience that would respect your exclusions while they may not be with original audiences.

But, again, I’m confident that the belief that exclusions aren’t respected with expanded audiences is a misinterpretation. When in doubt, go with the most likely explanation. And there are lots of them.

Your Turn

What are your feelings about the causes behind reaching excluded existing customers?

Let me know in the comments below!