Don't try this with MongoDB Folks

awesome.sql

  with v(dfrom, dto) as (
      VALUES('2014-01-01'::date, '2014-11-12'::date)
      --VALUES(_from, _to)
    ), failures as (
      select user_id, 'data-entry'::citext as action, count(*) as "count" from ad_histories, v 
      where action='qc-fail' and created_at>=v.dfrom and created_at<=v.dto 
      and exists(select 1 from ad_histories subq where ad_id=subq.ad_id and subq.action='data-entry' limit 1) group by user_id
    ), elapsed as (
      select r.user_id,r.action,avg(r.elapsed) as "avg_time" from (
        select user_id,action, lead(created_at,1,now()) over w - created_at as elapsed 
          from ad_histories,v 
          where created_at>=v.dfrom and created_at<=v.dto
          window w as (
            partition by user_id
            order by created_at
          )
      ) r group by r.user_id, r.action
    ), counts_by_status as (
      select ah.action,count(ah.*) as "count" from ad_histories ah,v 
      where created_at>=v.dfrom and created_at<=v.dto 
      group by ah.action
    ), counts_by_user as (
      select ah.user_id,ah.action,count(ah.*) as "count" from ad_histories ah,v 
      where created_at>=v.dfrom and created_at<=v.dto 
      group by ah.user_id,ah.action
    ), counts_with_failures as (
      select u.email,c.*,
        f.count as "failures",
        (select count from counts_by_status where action=c.action) as "total",
        ((c.count::float/(select count from counts_by_status where action=c.action)::float)*100)::int as "percent_count",
        ((f.count::float/c.count::float)*100)::int as "percent_failures",
        e.avg_time
      from counts_by_user c 
      join users u on u.id = c.user_id
      left join failures f on f.user_id = c.user_id and f.action = c.action  
      left join elapsed e on e.user_id = c.user_id and e.action = c.action
    ) select to_json(counts_with_failures) from counts_with_failures order by email,action;