The Problem: In Some Dialects, Performing a Calculation between Integers Always Returns an Integer

In certain dialects, including Postgres and Redshift, performing a calculation between integers (like dividing counts) will return an integer even if the result would be a decimal in normal math. You might make measures like this:

measure: sold_ratio {
 type: number 
 sql: ${sold_count} / ${total_count} ;;
 value_format: "0.00" # Number with exactly 2 decimals (1.23)
}

measure: sold_percent {
 type: number 
 sql: 100 * ${sold_count} / ${count} ;;
 value_format: "0.00" 
}

In the data table, however, the sold_ratio column is returning zero, and the sold_percent column does not have its decimal places populated. If we do the math, it becomes clear that this is not correct.

The Solution: Cast Field as a Floating-Point Number

If the calculation is multiplied by a non-integer, the values will cast as floats, and decimals will be returned as expected. You can multiply the numerator by a decimal number (like 1.0 or 100.0) to force SQL into returning a decimal result:

measure: sold_ratio {
 type: number 
 sql: 1.0 * ${sold_count} / ${total_count};;
 value_format: "0.00" 
}

measure: sold_percent {
 type: number
 sql: 100.0 * ${sold_count} / ${count};;
 value_format: "0.00" 
}