Usage
view: view_name {
  measure: field_name {
    approximate_threshold: 100000
  }
}
| 
       Hierarchy 
      
      
      
      approximate_threshold | 
    
       Possible Field Types 
      Measure
      Accepts 
      An integer
     | 
  
Definition
Google BigQuery Legacy SQL calculates an exact count_distinct if there are fewer than 1,000 distinct values, but performs an approximation for larger numbers of distinct values. The approximate count is faster, but typically has an error of a few percent.
If you want BigQuery Legacy SQL to calculate an exact count_distinct for more than 1,000 distinct values, you can set that amount with the approximate_threshold parameter.
Examples
Return an exact count_distinct for up to 100,000 distinct values:
measure: unique_count {
  type: count_distinct
  approximate_threshold: 100000   # default value is 1000
  sql: ${id} ;;
}