Glossary/Audience data/Predictive audiences
What are predictive audiences in advertising?
How predictive audiences are built, what they are used for, and their limits.
Written by Samhub with the help of AI·Updated 13 Oct 2026·3 min readPredictive audiences are groups of people defined by what a model predicts they will do, such as buy a car, move house or subscribe, based on patterns in past behavior.
How they are built
Find past examples
People who did the action before.
Learn the patterns
What they did in the weeks before.
Score everyone
How likely each person is to act.
Pick a threshold
Balance audience size and accuracy.
Common uses
16 pages for publishers, with a checklist for data, programmatic and direct sales.
Check the results
A prediction is only useful if it performs better than a broad audience. Measure the lift of each predictive audience before selling it.
Addressability at Stampen Media after moving its reader data into one layer with Samhub.
Frequently asked questions
It depends on the data and model. Test against a control group.
Usually, and GDPR applies.
Lookalikes resemble a seed group; predictive audiences forecast an action.
Regularly, as behavior and seasons change.
Related terms
New people who resemble your best audience.
Signals that someone is close to a purchase.
The share of subscribers who cancel.
Grouping readers so advertisers can reach them.
One EU-based platform to collect, unify and enrich your first-party data, installed with one script.
- A walkthrough with your own data
- Built for publishers and media houses
- 30 minutes, no commitment
Turn your audience data into more revenue
Get the Media Monetization Playbook: how media houses turn first-party data into higher CPMs, data deals and direct sales. With checklists for every step.
