The skinny on LinkedIn’s predictive audiences beta

Judging by the reaction in the Blackbird general Slack channel last week, there’s a lot of curiosity about LinkedIn’s predictive audience beta, which they are “gradually making available” (read: ask your agency to reach out to their rep).

In this post, I’ll break down the flurry of questions discussed in our Slack channel: what it is, what it isn’t, what you can use to enhance it, what LinkedIn says it might do for your performance, and a few more details that could pique your interest (or not).

What’s the difference between predictive and lookalike audiences?

Good question! Lookalikes are lists of audiences similar to whatever lists of contacts you upload, and they’re based on geo and demographics. Importantly, they have historically not performed all that well on LinkedIn across a few client tests.

Predictive audiences, meanwhile, are also based on lists of contacts, but they’re more precise: they’re people identified by LinkedIn as likely to take converting actions based on behavior, and they’re in similar positions and companies as the users on your list. In effect, this audience looks and acts like people you feed the platform.

What can you use to enhance it?

Along with a list of your most qualified converters, you can layer third-party segments on top of a predictive audience. We’re in the process of testing a layer of 6Sense segments, which give additional signals to signify in-market users. 

What could this do for your campaign performance?

LinkedIn says that predictive audiences are regularly showing CPLs 20% lower than other audience types. 

Do we believe this stat?

We’re half-hopeful, half-skeptical (see comment about poor lookalike performance above). But this is why we test.

What else do you need to know?

Per LinkedIn, make sure you’re aligning the data source with the converting behavior you’d like to achieve. It only takes 300 matches for LI to create a predictive audience, which should put the functionality within reach for a range of companies (assuming they have access).

A minimum of 300 list members - pretty achievable!

What’s next?

We’re excited about the tests we’re rolling out – and we’ll let you know when we’ve got some insight on performance to share. Stay tuned to the Blackbird blog!