We’ve updated how Faraday determines the number of personas in a set. By moving to a proportional scaling model, we ensure your segments are always optimized for your specific audience size.

### **What's new**

* **Dynamic scaling:** We replaced the "elbow method" with a proportional function (k=floor(log10​(cohort.size))).  
* **Utility-first clustering:** Persona counts now scale from 2 to 8 based on cohort volume, preventing over-segmentation of small lists and under-segmentation of large ones.  
* **Improved trait visibility:** Larger audiences now receive the granularity they deserve, making unique buyer traits more actionable.

This change is live for all new persona sets. Learn more in the [persona set documentation](/docs/abstractions/persona_sets).