Implicit conversions that blind your indexes

A predicate that looks selective can still scan when SQL Server must convert every row to compare types. Implicit conversions are a top reason “we have an index” still fails.

Typical causes

• VARCHAR column compared to NVARCHAR parameter
• Numbers stored as strings
• Mismatched precision/scale in joins
• ORM defaults that do not match table types

What reviews look for

• Plans with conversion operators on big joins/filters
• High CPU queries with unexpected scans
• Parameter types from apps that never matched the schema

Fixes

Align types at the schema and parameter boundary. Sometimes a computed column or careful indexing strategy helps; often the correct fix is discipline in the app contract.

Why it is blog-worthy for consulting

This is expertise clients feel immediately when fixed — and it is invisible without plan literacy. It is a natural follow-on from a health check that flags top CPU consumers.