Facebook Targeting: Lookalikes vs Pure Pixel

No matter what advertising platforms you use, how to best target your audience is always a hot debate. Different advertisers swear by different targeting methods and no one method has been proven for all industries. 

The endless amount of variables that can have an impact on performance make it impossible to determine what the X factor is. Maybe it was the product that made the difference at that time. Or it could also have the day of the week, supply and demand, or timing. There’s really no way to determine which factors had the impact. 

As you probably noticed, my posts don’t include academic case studies. They come from all the Facebook accounts I previously managed over the last seven years. Since day one, I’ve always constructed my campaigns in a systematic way that allows me to make comparisons between campaigns. From there, I can see what works across the board on hundreds of accounts and millions of dollars spent. While it’s not a clean a/b test, my large sample size helps me make a strong case.

If you read my articles about using interests to target audiences on Facebook, you’ll know that I advise against it. On the other hand, you may be wondering what I think about lookalike audiences. Using lookalikes is much more accurate than using interests because lookalikes are based on a larger list of similar people. 

Facebook’s definition of a lookalike audience is “a way to reach new people who are likely to be interested in your business because they’re similar to your best existing customers.”

Are lookalikes effective? Yes, but to a limited extent and during the early stages of your campaign. Let’s talk fundamentals for a second. The basic way campaigns are optimized is very similar to the way lookalikes work. Once your campaign starts getting conversions, a lookalike is created within each of your campaigns automatically, and targeting is shifted towards that audience. 

Even if you see that your audience size is large when you create your campaigns, you aren’t really targeting that entire amount. Once the campaigns start gathering conversions, a lookalike will automatically be created within the campaign and target a fraction of that audience. 

So the question is, should you use a lookalike source if your campaigns will already be creating lookalike models themselves?

Generally speaking, lookalikes will act as another layer of filtering on your existing audience. Lookalikes usually range from 200K-20M people in size depending on if you create them based on worldwide settings or individual countries. The drawback of using lookalikes is that your audience size will be limited. It will be hard to maintain a long-term, scalable campaign with a limited audience size. 

If your lookalike audience size is one million people to start with, think of how many of those people might actually be relevant to your business. Assuming the actual relevant audience is 10-20%, you are already looking at smaller numbers. Then, factor in the optimization process and you will find a pretty limited audience size.

The main issue with limited audience sizes is that they’re vulnerable to unscalable results. If your audience size is ten million or 30 million people, you are allowing the optimization algorithm freedom and flexibility to choose who to target and when so it will be able to provide steady, long term and scalable results. It will also give the optimization algorithm the ability to always look for the lowest cost opportunities across your audience instead of forcing it to deliver to a specific audience group where the costs might be high or demand might be low. 

Pure pixel is another option when it comes to targeting audiences on Facebook. The term pure pixel means using only your pixel data and campaign data for targeting optimization. 

Not using interests or lookalikes at all will allow the optimization algorithm complete freedom and flexibility in terms of audience and delivery. By doing this, your campaigns will only be optimized based on the conversion data in the specific campaign and from the data coming from the pixel on your website. 

The advantage of using pure pixel can be described in two words – steady scale. Using pure pixels opens you up to massive audiences and allows for steady performance at the lowest cost daily, allowing you to eventually scale your campaign to massive numbers. Once you have a campaign running and are able to achieve 20-50 conversions per day, targeting will automatically be set and you don’t need to add any layers of filtering. Think of how much data the campaign already has and allow the optimization to work freely for optimal results. 

Overall I’ve seen pure pixel consistently perform better than lookalikes across different industries and through different times. In some cases, ROI was very similar when comparing lookalikes to pure pixel, but pure pixel has always gotten much higher volumes. At times where lookalikes vs. pure pixel CPA and ROI were similar, pure pixel always delivered extra conversions and more reach. 

WHEN TO USE LOOKALIKES…

If you are not yet fully convinced of the benefits of using pure pixel method for targeting and would still like to try lookalikes, here is what I suggest:

1. Lookalikes should always be based on a list that represents your ultimate goal. For example, use a list of paying customer’s email addresses or a list of people who have completed a purchase based on your pixel. Do not create lookalikes of people who have simply visited your website or are not relevant to your business. 

2. Use rich information when uploading a list to create a lookalike list. Don’t just upload emails. Upload names, phone numbers and value per customer (lifetime value). In this way, Facebook will be able to create value lookalikes of not just paying customers, but top spenders. It will also help with better match rate and more accurate lists. 

You can start using lookalikes initially when your account is fairly new or doesn’t have conversions. As soon as you get close to the 50 conversions per week mark, start moving away from using lookalikes in new campaigns and move towards a pure pixel method to allow the optimization algorithm more freedom. 

IN SUMMARY

The effect of lookalikes is a much-discussed topic among marketers. Lookalikes are similar audience lists that are created based on a source like your customer list or email list. Lookalikes can limit your ability to scale your campaigns because they normally have much smaller audiences sizes. 

On the other hand, pure pixel is when you use only your pixel and campaign information to optimize your campaigns. This method allows you to achieve better performance across the board and scale your campaigns easily. Using pure pixel gives the optimization algorithm more freedom to provide more conversions for a better cost.