new facebook lookalike audiences

A New Approach to Facebook Lookalike Audiences Since the iOS 14.5 Update

If you advertise on Facebook you probably have heard or seen the affect that iOS 14.5 has had on advertising campaigns in Facebook.

The short version of what happened is that iOS 14.5 made it harder for Facebook to get information on what people are doing outside of the Facebook platform. This makes tracking less accurate and it makes it harder for Facebook to identify who is converting.

That means that since the iOS 14.5 update was released many advertisers are seeing results that aren’t as good as they once were and that lookalike audiences in particular aren’t performing as well as they used to.

We’ve seen a lot of the same kind of thing happening in the accounts we manage for our clients and we’ve come up with some adjustments and hacks that can mitigate some of the negative effects that folks are seeing from the iOS 14.5 update.

If you’ve seen a lot of drop off in your campaigns since iOS 14.5 rolled out and especially if your lookalike audiences aren’t working like they used to, then you want to make sure that you pay close attention to this guide. It could really help you out.

Why Lookalike Audiences Aren’t Working as Well Since the iOS 14.5 Update

We’ll start in my example Facebook ad account. This is an account I use for demonstration purposes and that’s why the numbers in here look like they are on the low side. I don’t actually run ads out of this account.

Sample Facebook ad account.

The issue that people are seeing with the lookalike audiences has to do with the source audiences. We know that the iOS 14.5 update has reduced Facebook’s ability to track user activity. When Facebook can’t track user activity as well, that reduces the amount of data that the algorithm has to work with and in turn reduces the effectiveness of a lot of the AI or machine learning that Facebook does.

Resource: If you don’t know how to set up lookalike audiences in Facebook, it’s something that you should understand and be able to do because it’s a very powerful way to increase your conversion rates. Please click here for “How to Create Lookalike Audiences” for a complete guide.

Lookalike audiences have traditionally been one of the best converting audiences both for my clients and in my own business. The good news is that they still can be great converting audiences!

The reason that lookalike audiences work so well is that Facebook has amazing machine learning that will take a “seed audience” or what it calls a custom audience and closely match that custom audience to other Facebook and Instagram users in order to help you get the best possible results. The algorithm literally goes out and looks for people that fit the same profile as the audience you’ve created as your custom audience.

The custom audience can be a lot of things. It might be an email list that you have, or it could be people who have purchased from you in the past, or it might be people who have watched more than 50% of one of your videos for example.

Resource: If you don’t know how to create a custom audience to make your lookalike audience from please click here for my post “How to Build Facebook Custom Audiences” and it will walk you step by step through the process.

For example, if you have a list of people who have bought from you in the past, you can upload that list to Facebook and use the lookalike audience to tell Facebook to go out and find other people with similar traits to those buyers. The people in your resulting lookalike audience are very likely going to be great people to advertise to and can improve your ROAS.

Using Larger Custom Audiences Has Become a Priority

In this example we’ll start with the custom audience section which you can find under the create audience button in your account.

Create custom audience in Facebook ad account.

Then we’ll do what worked best in the past and we’ll build my all time favorite source audience for our lookalike audience. That’s a custom audience that is made up of people who have triggered the Purchase event in my Facebook account. The default timeframe that Facebook gives you is 30 days, but you want to increase that to 180 days so you get more people in your source audience.

Set your timeframe to 180 days.

Historically, for most campaigns this has been the best performing source audience to build a lookalike audience from. But with iOS 14.5 things are changing. Anyone who has updated to iOS 14.5, which is most Apple users by now, isn’t getting tracked as they used to and that might mean they never make it into a source audience like this one.

The less data that Facebook has in your source audience, the less effective your lookalike audiences will be.

To a certain extent this has always been true. For example, if you have a buyers audience of 500 people and your competitor has a buyers audience of 5,000 people, then your competitor’s lookalike audience will almost always perform better than yours.

On the other hand, there is also a debate on what stage of the funnel is best to create your lookalike audience from.

People ask me all the time if it’s better to use a purchase event or an add to cart event as the custom audience source for their lookalike audiences. It’s a legitimate question because the purchase audience will be smaller and more precise, while the add to cart audience will be bigger but not as precise.

In the past I would have told people to test both but to expect the purchase audience to perform better.

But now, after the iOS 14.5 update, we often have to prioritize giving Facebook more data.

Now, if you were to test both lookalike audiences – one off of a smaller number of purchases and one off a larger number of add to carts, the add to cart audience might perform better.

Should You Go Even Higher in the Funnel?

The higher you go in the funnel, the bigger your custom audience can be. But the higher you go in the funnel the less like your buyers those folks will be as well. There’s a tradeoff here and testing is necessary.

In general, you want to be as close to your ideal prospect as possible WHILE giving Facebook the most data possible. Generally that means backing up a step in your funnel for your source audience.

As an example, if before the iOS 14.5 update you were using purchases as your source audience, I would now recommend testing a lookalike audience with add to carts as the source. You can try higher levels in the funnel like page views, but those tend not to be quite as good of a source audience to pull from.

Overall, by giving Facebook more data to work with, you give Facebook a good opportunity to improve your results.

Change Your Source to Make Your Lookalike Audiences Work Better

Upload your own list for a Facebook custom audience.

Another option that can work to make your lookalike audiences work better is to change the source data that you use.

Most of the time in the past I would use website events because those source audiences and the resulting lookalike audiences updated automatically and typically had high match rates.

Since iOS 14.5 you want to test uploading your customer list manually to Facebook. That means if you have a customer list from an email list or from your ecommerce store you should upload that list directly to Facebook to create your custom audience from. Facebook will then match the people in that list to Instagram and Facebook users. Because Facebook is now struggling to track website activity, non website sources, such as uploaded lists can result in a larger source audience.

Another BIG advantage to uploading these types of customer lists is that if your business has been around for a while then you can upload 3 ,4, or more years worth of buyers into Facebook. With Pixel data you can only go 180 days into the past.

Remember, bigger custom audiences are the priority! More data is better.

More Free Facebook Ads Training

To help business owners succeed with Facebook ads I’ve created a FREE webinar training that you can register for here.

Free Facebook ads training

When you attend this webinar you’ll learn:

  • 3 different Facebook ad strategies that we use every day. These strategies have generated millions of dollars in revenue and are tried and proven to work.
  • How to customize the Facebook ads strategy to your particular business. There is no such thing as a one size fits all approach to Facebook ads.
  • How Facebook and Instagram have changed and how to adjust your ad strategy to what works in 2021.

Video Tutorial: Changing Your Facebook Lookalike Strategy to Succeed With the iOS 14 Update

In this video I go over what’s changed with Facebook lookalike audiences since the iOS 14.5 update, why your lookalike audiences may not be performing as well as they used to and what to do about it.

The Bottom Line on Getting Your Facebook Lookalike Audience Working Again

The most important takeaway is that because of the iOS 14.5 update, the Facebook custom audiences that you build your lookalike audiences from, need to be bigger to give Facebook more data to work with. The primary methods we are using to get more data is to go up a level in the funnel for Pixel based audiences or to upload a customer list from your own database.

It’s also always a good idea to test different custom audiences for your lookalike audiences. You can test video views, lead form interaction, Facebook or Instagram page engagement and more. We’re seeing audiences from higher up the funnel perform better than they have before simply because the source audience is larger.

Whatever you can do to add more data into your targeting will be helpful.

We’re also finding that larger retargeting audiences are working well and broader targeting in our general campaigns is working well.

You want to test to see what works the best in YOUR business, and these are some great places to start.

  • October 22, 2021

    This may be a silly question, but it I am testing two Look-a-like audience ads, should the creative and content be the same for both. In my mind, they should be the same.

  • October 25, 2021

    Would you expect a 1% lookalike of page likes to be a good source? We have 26k+ page likes and the lookalike audience is approx. 500k

  • October 28, 2021

    Hello, When it comes to list size for creating a lookalike do you have a recommendation for how big that list should be bare minimum or ideally? Thank you so much :)!

Post a Comment