Should you combine or not combine lookalikes on Facebook?

By: Tom Murray | Managing Director

Over the past 3 years, I’ve audited at least 100 different Facebook accounts for both active and prospective clients, and one of the most common mistakes that I see is around lookalikes. Lookalikes are some of the strongest targeting options for finding new customers that are most likely to be similar to your current customers.

There are many sources that you can lookalike from. Here is just a sample of things you can lookalike off of:

1. Website custom audiences from pixel events
2. Video viewers
3. App installers
4. Email lists (purchasers, high value purchasers, leads, etc)
5. Page fans
6. Facebook & Instagram post engagers

What I normally see in accounts are a bunch of lookalikes created and then being run against each other to test if lookalike A does better than lookalike B. The problem with this approach is that by running all of these lookalike audiences individually, you are going to be a) overlapping audiences b) shrinking audiences and c) splitting out data across more ad sets.

The algorithm recommends 50 conversions per week in order to allow ad sets to exit the learning phase, which means combining ad sets into fewer groups consolidates data and helps ad sets learn faster. When testing lookalikes against each other, this data gets split into more and more ad sets, meaning you need to spend even more to get to the ideal learned phase.

Logically, when creating lookalikes it would make sense to split them out, knowing that a “high value” group is different than the “all purchaser” group. However, in the grand scheme of things in the Facebook ecosystem, the lookalikes that get created on both of those sources end up being very similar. We typically see audience overlaps of 70% or higher (and up to 95%) on 1% lookalikes of similar sources, and 50% on larger percentage lookalikes.

Instead of separating these out, creating “super lookalike” groups, which is when you combine multiple similar lookalike audiences into the same pool is our recommendation to give your ad sets the best chance at success.