Here's a basic thing that can save you a bit of time when analyzing data or validating data transformations.

So I've previously posted about how in BigQuery using LIMIT for query output does not yield any cost saving as it has no effect on amount on data being processed - just how many results are returned to you.

But there are still cases where I use LIMIT.

Say I'm validating some data and I want to check an assumption I have about the data. For instance, knowing that even a few duplicate records exist indicates me that the problem exists and provides an example to investigate.

I do not need to know all the possible duplicates in the table, therefore I use LIMIT to get at least one observation that will contradict what I'm expecting.

And even with LIMIT, if I don't get anything back, it means that the query hasn't found any matching rows which validates my initial hypothesis.

On a big enough table, one could notice the query execution time difference between using LIMIT and not using it. Again, there is no cost difference, but your time also costs 😁.

P.S. This is not to say that LIMIT is completely irrelevant to performance in BigQuery. Check out this post for a case where LIMIT does make a difference!

Found it useful? Subscribe to my Analytics newsletter at notjustsql.com.