Edit Thresholds for Classification Prediction Models in “Published” Mode 

You can now edit the “True/False” threshold of a Classification Prediction Model that is already in “Published” mode — without retraining it. 

What’s new? 

Until now, adjusting the threshold was only possible right after training, before publishing the model. Starting today, you can revisit any “Published” model and fine-tune the threshold based on its actual results — dynamically deciding the cutoff point above which the model predicts “True.” 

Why it matters: 

This enhancement allows our clients to:     

• Instantly adjust prediction logic without starting from scratch
• Refine audience definitions and activations with precision
• Make their customer experience even more adaptive to changing business goals

Take Me There