Skip to main content
After your lookalike model has finished training, you can use the resulting curve to create specific cohorts for targeting.

​
Understanding the Curve

The Precision vs. Reach curve represents the trade-off between how closely users match your seed (Precision) and the total number of users you can target (Reach).
  • High Precision (e.g., 90%): Users match the seed behavior very closely, but the audience size will be smaller.
  • Low Precision (e.g., 10%): A much larger audience, but with a weaker behavioral match to the seed.
1
Open Your Model
2
Navigate to Modeled Cohorts > Models and click on your trained lookalike model.
3
Select a Point on the Curve
4
Click any point (represented by a circle) on the curve that matches your desired balance of reach and precision.
5
Configure the Cohort
6
A pop-up will appear. Provide the following details:
7
  • Name: Give your lookalike cohort a descriptive name (e.g., “Tennis Lovers - 80% Lookalike”).
  • Activation Type: Choose how you want to use this cohort:
    • Publisher Ad Server: For direct campaigns.
    • Programmatic: For activation via SSPs and DSPs.
    • 8
    Save and Activate
    9
    Click Create Cohort. Your new lookalike cohort will now appear in your list of cohorts and is ready for activation.
    You can create multiple cohorts from the same model at different similarity levels to test performance across different audience sizes.

    ​
    Important Notes

    • 100% Similarity: Choosing 100% similarity will result in an audience identical to your seed segment with no lookalike expansion.
    • Real-time Scoring: Once created, users are scored against your lookalike cohorts in real-time as they browse your site.