Update Hard-Coded References in Unstructured Data Queries

On December 8th, 2025 - Salesforce is implementing a performance update for the Query endpoints below that may impact your Data 360 (formerly Data Cloud) application if you have hardcoded column names or data types in your Unstructured SQL query flows. While the service's core functionality will remain unchanged, Salesforce is updating some column metadata and column naming conventions in the data output. We recommend that you review your client code that issues SQL queries that use vector_search or hybrid_search functions to ensure that it does not rely on hardcoded column names or data types.

Important: Due to performance optimizations, unstructured SQL queries will return slightly different output metadata. Customers who depend on column names or data types during the migration period may experience breaking changes.

Impacted Endpoints:

• Connect API Query API v1
• Connect API Query API v2
• Direct API Query API v1
• Direct API Query API v2

Data queries that use the Data 360 Query v1 and Query v2 Connect & Direct API endpoints along with the vector_search or hybrid_search functions  will be impacted in these scenarios:

1. Dependencies on Column name Casing for "select *" queries
   1. Impact: Instead of lowercase column names, applications will receive original API name casing from metadata
   2. Example: Code that looks for ssot__id__c will receive ssot__Id__c
   3. Recommended Action: Implement case-insensitive column name matching, or use column aliases in your queries

2. Dependencies on Timestamp field metadata format

   1. Impact: Applications expecting `TIMESTAMP WITH TIME ZONE` will receive `TIMESTAMP_WITH_TIMEZONE`
   2. Example: Code that checks `if (type == "TIMESTAMP WITH TIME ZONE")` will fail
   3. Recommended Action: Update timestamp type validation to handle both formats

Data queries that use the Query v2 Connect & Direct API endpoints along with the vector_search or hybrid_search functions will in addition be impacted by these scenarios as well:

1. Dependencies on Decimal Handling
   1. Impact: Applications expecting decimal values and `decimal` type will receive string-wrapped values with `numeric` type
   2. Example: Code that expects 123.45 (decimal) will receive "123.45" (string)
   3. Recommended Action: Update your code to handle both decimal and string-wrapped numeric value.

Data queries that use the Query v1 Connect & Direct API endpoints along with the vector_search or hybrid_search functions will in addition be impacted by these scenarios as well:

1. Dependencies on column name for Aggregate Function calls

   1. Impact: Column names will change from aggregate-specific names like `count` to generic names like `col_0`
   2. Example: Code that references `result.getColumn("count")` will fail
   3. Recommended Action: Use column aliases in your queries (e.g., `SELECT COUNT(*) as count_column`), or implement column position-based access
2. Dependencies on Decimal type metadata
   1. Impact: Applications expecting `decimal` type metadata will receive `numeric` type
   2. Example: Schema validation that checks `if (fieldType == "decimal")` will fail 
   3. Recommended Action: Modify type checking to accept both `decimal` and `numeric` types, or implement a type mapping function