Counting results that have two different values in the same column

Question

I work in support and I am trying to count how many 'cases' have both of these labels applied:

'Chat' and 'Escalated'

This is the query I am using, but I am getting an incorrect number:

    select count(pk_case_number)
    from cases c
    join labels l
    on l.case_number = c.case_number
    where lower(label_name) = 'chat' 
    and exists (select distinct case_number
    from cases c
    join labels l
    on on l.case_number = c.case_number
    where l.case_number = c.case_number
    and lower(label_name) = 'escalated')
    and date(created_at) > '2013-08-11' 

'created_at' is in the cases table.

Answer 1

The following query lists all the cases where the condition is true:

select pk_case_number
from cases c join
     labels l
     on l.case_number = c.case_number
group by pk_case_number
having sum(lower(label_name) = 'chat') > 0 and
       sum(lower(label_name) = 'escalated') > 0;

The following counts them:

select count(*)
from (select pk_case_number
      from cases c join
           labels l
           on l.case_number = c.case_number
      group by pk_case_number
      having sum(lower(label_name) = 'chat') > 0 and
             sum(lower(label_name) = 'escalated') > 0
     ) t

I like doing these sorts of queries with aggregation and the having clause, because I find that to be the most general. It is easy to add new labels that are required or to exclude a particular label (use = 0 rather than > 0 ).