In what order should you set columns for a multicluster key?

• A. Highest cardinality to lowest cardinality
• B. Lowest cardinality to highest cardinality
• C. Alphabetical order by column name
• D. Order does not matter

Answer: (B)

In a multicluster key, the order of columns defines how data is distributed across clusters and how pruning happens. The first column serves as the broadest grouping, so you want it to have the fewest distinct values. Placing the lowest-cardinality column first creates a manageable set of clusters and lets Snowflake prune data more effectively when queries filter on that column. The higher-cardinality columns come next to refine within those clusters, providing detailed discrimination only where needed.

If you put a high-cardinality column first, you’d effectively create many more clusters, which can increase overhead and scatter data, making pruning less efficient. Alphabetical order or assuming order doesn’t matter won’t reliably align with how data is actually distributed or how pruning works, so they aren’t good guidelines for multicluster keys.