Collecting statistics on data in Snowflake mainly serves to enable efficient pruning based on query filters. These statistics inform the query optimizer about the data distribution and characteristics, allowing it to skip irrelevant micro-partitions and thereby improving query performance by reducing the amount of data that needs to be processed.