menu close
menu close
back to Insight Page

Nested Lookalike Audiences: How To Avoid Two Major Digital Advertising Pitfalls

Written by: Juliana Casale, Director of Marketing

A few months ago, we posted about the powerful effects of pairing Facebook’s Custom Audiences (people who have expressed interest in your brand) with Lookalike Audiences (people who behave similarly to the aforementioned cohort). Our direct response customers are seeing a lot of success with this combination, and it comes highly recommended as a best practice from our account managers. However, there is one major caveat to the approach that all digital marketers should be aware of: Cannibalization.

Say you’re targeting a 1% Lookalike Audience based on people who have browsed bicycles on your website, and it’s doing really well. You want more reach, so you generate a 1% Lookalike Audience based on bicycle helmet browsers. Overlap between those two groups is likely very high, in which case you’re hitting the same cohort twice. This scenario can lead to two advertising no-no’s:

1. Oversaturating your customers with ads, and

2. Bidding against yourself.

How you can avoid cannibalizing your best customers

Lookalike audience nesting

Audience saturation via Lookalike Audiences means people similar to your best customers are being hammered with ads. You may lose their attention, and you will certainly lose money. For brands with products (and ads) that don’t change seasonally, it can be especially easy to overexpose prospects in this manner.

Enter Nesting Lookalikes – a strategy for expanding your Lookalikes beyond 1% that can be used as an alternative to creating more seed audiences for Lookalikes.

The concept is very similar to Russian nesting dolls. You take an audience that represents the 1% of the US population that looks most like your best customers, exclude it from the next group of people that are slightly less similar (2% or 3%), and so on.

As you move from ring to ring, you exclude the segment before it. Then you test each audience to see where the best performance is. Typically, 1% shows the most promising returns, but sometimes our customers are surprised by the results; the 1% cohort is okay, while 3% is the best.

In Nanigans, nested Lookalikes are easy to execute and only take a few minutes to set up.

The process is simple, but its value is profound. Without nesting Lookalikes, digital marketers are unable to determine which cohort is driving performance. They are wasting spend showing people ads they’ve seen before, and creating unnecessary campaign sprawl.

Best practices concerning Lookalike Audiences

Lookalike best practicesWhen working with nested Lookalikes, it’s important to be cognizant that most/all of the best practices concerning Lookalikes still apply:

  • The seed audience should be at least 5,000 users, but we recommend using at least 30,000 users. This helps ensure there’s sufficient data to find the best Lookalikes.
  • When you optimize for similarity, Facebook will find the people that are most similar to your existing audience. The estimated reach of the new audience will be smaller, but the match will be more precise. When you optimize for reach, Facebook will find more people who are similar but may be a less precise match to your audience.
  • Your seed audience should be composed of people who have taken the action you’re trying to drive. If you are trying to drive purchases of a particular product, then upload an audience made up of people who have bought that or similar products in the past. In the bicycle example we cited at the beginning of this post, it might be worth combining bicycle and bicycle helmet seed audiences to make a better informed Lookalike Audience from the beginning and then use the nested Lookalikes strategy.
next post

But wait, there's more

Join Our Newsletter