Recommended amount of columns OR expressions per key

• A. 1 or 2 columns
• B. 3 or 4 columns/expressions
• C. 5 or 6 columns
• D. 7 or more columns

Answer: (B)

In Snowflake, a clustering key determines how data is organized into micro-partitions to improve pruning during queries. The goal is to balance pruning effectiveness with the overhead of maintaining clustering metadata. Keeping the clustering key small—three to four columns or expressions—provides a practical sweet spot: it captures common filter patterns without incurring the diminishing returns and extra maintenance that come with many more columns. You can include expressions (for example, a date_trunc expression) if they align with how you query, but they still count toward that limit and should be used sparingly. If you choose too few, you may miss opportunities to prune; if you choose too many, the overhead can outweigh the benefits.