Maximising Campaign Success with LinkedIn’s Predictive Audiences: An AI-Driven Approach

By Space & Time
13 Jun 2024

With the rise of AI technologies through tools such as ChatGPT, we have seen an Social Media platforms adapt their strategies to encompass this new capability to improve performance, whether it be Meta’s Advantage+ or TikTok’s creative assistant. LinkedIn have delivered their own version of an AI tool through predictive audiences.

Predictive audience’s combine data from an existing source with LinkedIn’s AI to generate a new custom audience of people who would be more likely to convert. This data source has to have at least 300 people and can be a customer list, lead gen form or people who have converted at a specific conversion point. You can only use one type of data source per predictive audience but can combine multiple versions of that source to ensure you have 300 people such as selecting multiple lead gen forms. Check out our Drum article here for more information.

Our insights: A look at a live campaign

We set up a predictive audience using a customer list of people who had previously converted at the conversions lowest in the funnel so LinkedIn would generate a custom audience with the highest intent. We found that the first month of activity is crucial for initial learnings and led to an 333% increase in conversions in the second month of activity as well as a 32% decrease in CPM, despite a 43% decrease in ad spend.

From LinkedIn’s testing they saw a 21% decrease in cost per lead (CPL) when using predictive audiences and lead gen objectives and we saw a 38% decrease in CPL from the second month of predictive audience’s being used compared with standard targeting. From this activity our recommendation was to run a predictive audience campaign for at least 3 months to ensure efficient learnings as performance was projected to improve MoM.

This performance indicates the importance of data driven targeting  and embracing new technologies. As each platform evolves to offer improved targeting capabilities, we should adopt a test and learn strategy to assess the best use for the newest technology to deliver optimal performance. Another factor that should be considered when using predictive audiences is strong creative; the creative should have clear messaging to insight the desired action based on the characteristics of the data source used for the predictive audience.

AdvertisingArtificial Intelligence & Machine LearningClient servicesContentDataGrowthStrategy

Latest news