How LinkedIn’s predictive audiences give B2B marketers the competitive edge

AI, even with the limitations of any innovation during its infancy, is already shaping the media channels we encounter daily. Just look at LinkedIn’s most recent predictive audiences (PA) update, intended to support B2B advertisers’ efforts to increase conversion volume.

This optimization AI technology is set to give some a competitive advantage, aiming to increase the number of conversions and reduce cost per goal.

As with any innovation of this sort, a structured and rigorous approach to testing is necessary. Caution aside, though, it does seem likely that this creates a viable targeting option for many businesses’ Q4 strategies and beyond.

Predictive audiences: The basics

Often featuring bigger-ticket sales, longer purchase cycles and lower transaction volumes than B2C activity, B2B typically has fewer data points for algorithms and AI to work with than B2C. So understanding the value of this latest tech within trade verticals will be particularly useful.

LinkedIn’s predictive audiences uses AI to create new audience segments based on first-party data, lead gen form completions, or data built from conversions that take place on the website. The social network’s AI model extracts demographics, firmographics (information about organizations) and behavioral attributes to predict the users most likely to exhibit similar conversion behavior.

This audience type can be compared to lookalike audiences, as it uses data including website audiences, video ad audiences, company lists, and contact lists to build a lookalike. But while lookalikes use a fixed methodology to find a new audience, PA uses a multitude of data points, learning and evolving as campaigns run. As more data becomes available over time, the technology powering these audiences can make predictions based on more indirect and granular insights.

What insights do we have so far?

LinkedIn recommends that this audience type is best for lead generation, as it identifies buyers that have high intent.

In its own testing phase, LinkedIn saw a 21% cost-per-lead reduction in campaigns using predictive audiences and lead gen objectives. If this is replicable, this new audience type will improve ROI for businesses and create targeting that the B2B world has never seen before.

Predictive analytics: How to get started

When using PA, make sure that the data you use to seed the audience is high-quality, recent, larger than 300 records, and has generated a high-value conversion. Lower-funnel conversions should not be used.

Use no more than 200x your seed source size when creating the audience, and be sure to align your content to the audience in order to increase the likelihood of conversion.

Then, it’s important to keep your goals in mind when creating and targeting your predictive audience. Make sure that your goals are specific and measurable and that you’re targeting the right users with the right messaging. Run the campaign for at least three months, allowing time to record learnings and feed them back into the AI algorithm for it to evolve. You can then run tests with formats, creative, and against other campaigns.

Data-driven targeting is the key to a successful marketing campaign; it's especially important when using predictive audiences. The key takeaway is nothing new within the broader marketing context: the value of using data to inform your targeting decisions cannot be overstated. The other key here is agility: be ready to adjust your targeting strategy based on the data you collect, and embrace technology that can do that at scale and speed.