What types of columns are most useful when selecting for clustering?

• A. Actively used columns in selective filters
• B. Columns with highest data type size
• C. Columns with constant values
• D. Columns with random distribution

Answer: (A)

Clustering improves performance when you organize data by columns that are actually used to filter queries. If a column frequently appears in WHERE clauses or other predicates, clustering on that column helps Snowflake prune micro-partitions that don’t match the filter, so only relevant data is scanned. This reduces compute and speeds up responses.

The other options don’t help with pruning: choosing columns based on data type size doesn’t affect how partitions are pruned; a column with constant values offers no selective advantage, since all rows would share the same value; and a column with random distribution won’t yield predictable pruning, making clustering far less effective.