Yes, we know that audiences aren’t sexy anymore.

We also know that, <low-highlight>post-iOS14, most advertisers strongly favor open targeting or DABA campaigns<low-highlight>.

<low-highlight>However, we are seeing Lookalikes perform very well<low-highlight> in many of our accounts. In this post, I am going to break down why these audiences are working, when so many others aren’t.

One thing you need to know before you start — these suggestions may be more useful for brands with a large number of past purchasers (think 50K+, but 100K is ideal).

Okay, let’s get down to it.

Without data, all you have is opinion

If you have <blue-low-highlight>more than 50,000 past purchasers<blue-low-highlight>, you’re in luck.

Not just because you’re selling a lot, but because you have enough first-party data (data that belongs to you) to make effective Lookalikes.

Why does this matter?

We’ve covered this in more detail here but, briefly, post-iOS14, Meta can no longer capture enough data from its pixel to know:

• who your most valuable buyers are, 
• when they're in the market for something new, and 
• how much they are looking to spend.

This lack of data affects Meta’s ability to target your ads with the same effectiveness that it did before iOS14. 

However, the good news is that you have enough data of your own to send Meta highly relevant signals. <blue-low-highlight>Your past purchaser data is valuable information that can help Meta predict new buyers for your brand.<blue-low-highlight> 

The most accurate persona for your products are those who have already bought those products.

Doubters gonna doubt

We’ve heard a lot of people say that Lookalikes don’t work anymore.

We disagree.

Typically, we have found that the naysayers fall under one of the following two categories:

• They are making Lookalikes (LALs) incorrectly i.e. using Klaviyo. Learn more.
• Their existing CAC (Customer Acquisition Cost) is very high indicative of other issues in the business. 

LALs  aren’t a magic pill. However, when made correctly (learn how here), LALs are still working well for many of our clients. 

But to take your LAL performance to the next level, you need to add another layer — segmentation.

The next step - break it down

Anyone with past purchase data can send it to Meta. But with more data, comes greater accuracy.

You have enough information to <blue-low-highlight>create potent, data-driven audience segments<blue-low-highlight>.

Basically, this involves dividing up your audience into ‘buckets’ based on similarity of behavior, demographics, or psycholographics. 

Audience segments that you may already be familiar with include repeat buyers, high spenders, or recent shoppers. 

At Socioh, we recommend two segmentation strategies: engagement-based segmentation and persona-based segmentation.

Engagement-based segmentation

Inside Socioh, all your past performers are auto-grouped into segments using machine learning.  

Our recommendation is to only take your top 3 segments and make a custom audience from them: a one-click operation through Socioh. You most valuable customer segments, they are ideal for making LALs.

This is a great way to identify profitable audience segments hidden in your own data.

Think of this as putting your best foot forward. 

Meta gets only your best customers to learn from, which significantly improves the quality of the signal being sent to Meta.

Persona-based segmentation

“Segmentation is a natural result of the vast differences among people.” - Donald Norman

People are different. But effective marketing needs to be personal.

So how do we make sure that we are communicating the best possible message to each potential customer? How do we make each shopper feel like we are talking just to her?

Through persona-based segmentation.

So how do you use your purchaser data to make data driven personas?

At Socioh, one of the ways you can do this is by <blue-low-highlight>creating segments based on the category that your buyer has purchased from<blue-low-highlight>.

Let’s say you sell sports equipment. You probably stock everything from tennis balls to swimwear to exercise bikes. In your store, you wouldn’t talk to a shopper for a golf bag the same way that you would to someone interested in yoga mats - you know that would just cost you a sale.

And yet, we do this all the time in ads.

Advertisers are making LALs based on all past purchasers, letting Meta target all shoppers interested in sports with what may be category-specific creatives.

Instead, what if LALs using your golf-category buyers were shown creatives made for golfers while the LALs based on your yoga buyers were shown ads for yoga wear? Wouldn’t that make more sense?

If you are talking to everyone, you are talking to no one.

Conclusion

To recap, Lookalike Audiences can still work if done correctly. <low-highlight>One reason why Lookalikes are not performing well is because advertisers are creating them incorrectly through email services like Klaviyo<low-highlight>.

Many of the brands working with Socioh are seeing great results from Lookalikes by using RFM-based audiences and then segmenting this audience further. At Socioh, the two types of audience segmentation that we are seeing most success with are (a) Engagement-based Segmentation, and (b) Persona-based Segmentation

We’d love to get your thoughts on how these tips worked out for you.