When defining a multi-column clustering key, which ordering is generally recommended?

• A. From lowest cardinality to highest cardinality.
• B. From highest cardinality to lowest cardinality.
• C. The order does not matter.
• D. Always place highest cardinality first.

Answer: (A)

Clustering keys determine how Snowflake lays out data in micro-partitions and how pruning happens when you run queries. When you define multiple columns in a clustering key, the first column’s values guide the top-level partitioning, and the following columns refine the pruning inside those partitions. Placing the column with the fewest distinct values first creates broader, more repeatable groupings at the top level, which helps Snowflake quickly exclude large chunks of data that don’t match the filter. Then, within the selected groups, the higher-cardinality columns can narrow the search even further. If you start with a high-cardinality column, the initial pruning is less effective because there are many distinct values across partitions, so more partitions may be touched before the subsequent columns can prune, reducing overall efficiency. So ordering from lowest cardinality to highest cardinality generally yields better partition pruning across common query patterns.