How Facebook Lookalike Targeting Works
Facebook lookalike audience targeting is one of the most powerful tools that mobile UA marketers have at their disposal. It is not, however, without its limits.
Lookalike audiences work by taking an existing dataset (Facebook recommends between 1,000 and 50,000 people), and finding other similar users based on certain qualities and behavior. To define the size of your target audience using this technique, Facebook uses percentage of the entire Facebook population in your target geography. So if you’re targeting users in the U.S., a country of about 300 million people at 1% lookalike audience your addressable audience is about 3 million. At 2% it’s 6 million and at 5% it’s 15 million.
Sounds like a good deal, right?
Relative to other ad products on the market, it is great. You are able to quickly scale your addressable audience with reasonable confidence that your are reaching the right people. The challenge, specifically for app marketers, comes in two forms:
- The quality of audience decreases as your scale from 1%
- There is an opportunity cost associated with failing to target your complete audience from the start
Why CPI Goes Up After 1%
At 1% even affinity-based targeting can only match the efficacy of Facebook lookalike audiences. But if you find yourself in a position where either your conversion rate is lower than expected or you need to continue your rapid growth beyond what you can get with 1% lookalikes, you might run into trouble.
As the percentage increases with lookalike audience, the quality decreases. This is because the correlation between your seed set of users and people that Facebook deems similar get weaker as the platform tries to add in more and more people. For purely illustrative purposes, if 1 in 5 people in your 1% lookalike audience are truly relevant users that are likely to download and use your app, at 2% maybe it drops to 1 in 8, at 5% maybe it’s 1 in 50 and so on.
This creates big problems for app marketers with increasingly rigid demands for ROI. Once you hit the boundaries of 1%, your back is already against the wall and the fight for targeting quality users without hurting your CPI becomes more difficult. Once you realize you need a new strategy, it’s too late.
You’re spending more than you should, even within 1%
Even if you never scale beyond the 1% audience and are able to hit your growth targets without surpassing your CPI goals, you could be spending more than necessary. In the 2% and beyond audience segments, there may be fewer relevant users for every 100 impressions Facebook can deliver, but those users are still there and still being missed.
If you aren’t targeting your complete audience at the start of your campaign, your bids are not optimized as efficiently as they would be otherwise. By targeting a smaller audience, you are more likely to bid more for an impression in instances when you could acquire another user in another segment (2% and beyond) for much less.
Even if you are below your CPI goal, wouldn’t you accept acquiring those same users and others like them for less?
What you can do about it
The first step is to target your complete audience from the beginning. In addition to lookalike audiences, consider how you can expand your interest targeting beyond the obvious.
Appnique has a suite of tools that help you build high affinity audiences to complement your lookalike campaigns so that your cost-to-acquire doesn't go up as you extend your reach. By mining app store data to identify audience segments that are likely to download and use your app, Appnique is able to map them back to Facebook interests for easy targeting to keep your acquisition efforts healthy even as you scale.
To get started, check out our free trial or sign up for a 10-15 minute consultation where we will analyze your app's install base or that of a competitor.