Exact Data Match protects structured sensitive data by converting source values into secure hashes before they are used by Zscaler cloud enforcement. The on-premises EDM VM performs indexing locally so raw customer data is not uploaded into the Zscaler cloud. Only hashed values are used for matching. Option B (By using an on-premises VM to index data and only sending hashed values to the cloud) is correct because the VM automates indexing and sends hashed data, not clear sensitive records, to the cloud.

Why the other options are incorrect:

A. By storing sensitive structured data on servers managed by trusted Zscaler staff for enhanced security: Storing raw structured data on Zscaler-managed servers would create unnecessary exposure. EDM avoids this by sending hashed values, not the original data.

C. By requiring customers to manually hash the data and upload it to the cloud: Manual customer hashing is error-prone and not the EDM workflow. The on-premises VM automates hashing/indexing before values are sent to Zscaler.

D. By encrypting sensitive data directly before storing it in the cloud: Encrypting raw sensitive data before cloud storage still implies the data is being stored. EDM avoids that by sending hashes, not the original structured values.