CASE expression from multiple tables - sql

Question

I'm trying to write a CASE expression where I look at 2 tables (same DB) to output a 1 or 0 and then SUM that column with some grouping on BRANCH and EVUSERID.
This may be the wrong method altogether for what I'm wanting to achieve so am completely open to suggestions.
EVENT table
+--------+----------+-----------+---------------------------+-----------+
| BRANCH | EVUSERID | EVENTTYPE | REFERENC | ENQUIRYNO |
+--------+----------+-----------+---------------------------+-----------+
| 1 | cunnc | SAS | New Cust - Unknown Source | 1468731 |
| 2 | malce | SAS | New Cust - Walk-in | 1792469 |
| 1 | cunnc | SAS | Old Cust - Retention | 897324 |
| 4 | wardt | OPT | | 986234 |
| 5 | shera | OPT | | 576324 |
| 4 | hamzk | SAS | Old Cust - Service | 1238746 |
| 5 | chinc | SAS | New Cust - Digital | 687236 |
+--------+----------+-----------+---------------------------+-----------+
SORDER table
+-----------+---------+
| ENQUIRYNO | NEWUSED |
+-----------+---------+
| 1468731 | N |
| 1792469 | N |
| 897324 | U |
| 986234 | N |
| 576324 | U |
| 1238746 | U |
| 687236 | U |
+-----------+---------+
My desired output would be
+--------+----------+----------------+---------+-------+---------+-----------+---------+-------------+--------------+---------+
| BRANCH | EVUSERID | Unknown Source | Walk-in | Phone | Digital | Retention | Service | New Enquiry | Used Enquiry | Unknown |
+--------+----------+----------------+---------+-------+---------+-----------+---------+-------------+--------------+---------+
| 1 | cunnc | 1 | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 0 |
| 2 | malce | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 0 | 0 |
| 4 | hamzk | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 1 | 0 |
| 5 | chinc | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 0 |
+--------+----------+----------------+---------+-------+---------+-----------+---------+-------------+--------------+---------+
Where I am so far. All is working as expected until I added my join in
SELECT
E.CM_EV_BRANCH,
E.CM_EV_EVUSERID,
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND E.CM_EV_REFERENC LIKE '%Unknown%' THEN 1 END) AS 'UnknownSource',
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND E.CM_EV_REFERENC LIKE '%Walk%' THEN 1 END) AS 'Walk-in',
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND E.CM_EV_REFERENC LIKE '%Phone%' THEN 1 END) AS 'Phone',
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND E.CM_EV_REFERENC LIKE '%Digital%' THEN 1 END) AS 'Digital',
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND E.CM_EV_REFERENC LIKE '%Retention%' THEN 1 END) AS 'Retention',
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND E.CM_EV_REFERENC LIKE '%Service%' THEN 1 END) AS 'Service',
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND S.VM_SO_NEWUSED = 'N' THEN 1 END) AS 'New Enquiry',
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND S.VM_SO_NEWUSED = 'U' THEN 1 END) AS 'Used Enquiry',
SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND S.VM_SO_NEWUSED <> 'U' AND S.VM_SO_NEWUSED <> 'N' THEN 1 END) AS 'Unknown'
FROM
Datalake.dbo.CM_01_EVENT E
LEFT JOIN Datalake.dbo.VM_01_SORDER S ON E.CM_EV_ENQUIRYNO = S.VM_SO_ENQUIRYNO
WHERE
E.CM_EV_BRANCH <> 0 AND E.CM_EV_DEPTTYPE = 'S'
GROUP BY
E.CM_EV_BRANCH, E.CM_EV_EVUSERID
/*
HAVING SUM(CASE WHEN E.CM_EV_EVENTTYPE = 'SAS' AND E.CM_EV_REFERENC LIKE '%Unknown%' OR E.CM_EV_REFERENC LIKE '%Walk%' OR E.CM_EV_REFERENC LIKE '%Phone%' OR E.CM_EV_REFERENC LIKE '%Digital%' THEN 1 ELSE 0 END) <> 0
*/
ORDER BY
E.CM_EV_BRANCH, E.CM_EV_EVUSERID
I've commented out the HAVING clause as that didn't work as desired. There are instances in the DB where an EVUSERID exists for a given branch but doesn't meet the criteria in the CASE expressions. These would display as "0" which whilst is correct, I wasn't interested in seeing during developing the SQL statement.
Any help would be gratefully appreciated.

Answer 1

I think you want a where clause: this filters out rows that would have all 0s, and lets you simplify the case expressions in the subquery (since one condition is common to all expressions).
Also: don't use single quotes for column aliases! They are meant for literal strings. Use aliases that do not require quoting - or use square brackets to quote them.
So:
SELECT
E.CM_EV_BRANCH,
E.CM_EV_EVUSERID,
SUM(CASE WHEN E.CM_EV_REFERENC LIKE '%Unknown%' THEN 1 ELSE 0 END) AS UnknownSource,
SUM(CASE WHEN E.CM_EV_REFERENC LIKE '%Walk%' THEN 1 ELSE 0 END) AS Walk_in,
SUM(CASE WHEN E.CM_EV_REFERENC LIKE '%Phone%' THEN 1 ELSE 0 END) AS Phone,
SUM(CASE WHEN E.CM_EV_REFERENC LIKE '%Digital%' THEN 1 ELSE 0 END) AS Digital,
SUM(CASE WHEN E.CM_EV_REFERENC LIKE '%Retention%' THEN 1 ELSE 0 END) AS Retention,
SUM(CASE WHEN E.CM_EV_REFERENC LIKE '%Service%' THEN 1 ELSE 0 END) AS Service,
SUM(CASE WHEN S.VM_SO_NEWUSED = 'N' THEN 1 ELSE 0 END) AS NewEnquiry,
SUM(CASE WHEN S.VM_SO_NEWUSED = 'U' THEN 1 ELSE 0 END) AS Used Enquiry,
SUM(S.VM_SO_NEWUSED NOT IN ('U', 'N') THEN 1 ELSE 0 END) AS Unknown
FROMDatalake.dbo.CM_01_EVENT E
LEFT JOIN Datalake.dbo.VM_01_SORDER S ON E.CM_EV_ENQUIRYNO = S.VM_SO_ENQUIRYNO
WHERE
E.CM_EV_BRANCH <> 0
AND E.CM_EV_DEPTTYPE = 'S'
AND E.CM_EV_EVENTTYPE = 'SAS'
AND (
E.CM_EV_REFERENC LIKE '%Unknown%'
OR E.CM_EV_REFERENC LIKE '%Walk%'
OR E.CM_EV_REFERENC LIKE '%Phone%'
OR E.CM_EV_REFERENC LIKE '%Digital%'
OR E.CM_EV_REFERENC LIKE '%Retention%'
OR E.CM_EV_REFERENC LIKE '%Service%'
)
GROUP BY E.CM_EV_BRANCH, E.CM_EV_EVUSERID
ORDER BY E.CM_EV_BRANCH, E.CM_EV_EVUSERID