SQL statement using WHERE from a GROUP or RANK (part 2) - sql

Question

I recently posted a question about a SQL Where Statement/Grouping here:
SQL statement using WHERE from a GROUP or RANK
Now I've got somewhat of a follow-up.
So similar to the previous question, let's assume I have a table of say 35,000 rows with these columns:
Sales Rep | Parent Account ID| Account ID | Total Contract Value | Date
Each row is individual by account id but multiple account IDs can fall under a parent account ID.
Similar to the responses on the first question, this is probably going to be a table w/i a table. So first, everything has to be grouped by Sales Rep. From that, everything needs to be grouped by Parent Account ID where the grouped total contract value of all the accounts is >= 10,000. Then everything will be displayed and ranked by the total TCV of the Parent account ID and I need the top 35 Parent account IDs by agent.
So the first couple of lines of data may look like this:
Sales Rep | Parent Account ID| Account ID | Total Contract Value | Date | Rank
John Doe | ParentABC12345 | ABC425 | 5,000 | 1/2/2013 |1
John Doe | ParentABC12345 | ABC426 | 10,000 | 1/2/2013 |1
John Doe | ParentDJE12345 | DJE523 | 11,000 | 1/2/2013 |2
John Doe | ParentFBC12345 | FBC6723 | 4,000 | 1/2/2013 |3
John Doe | ParentFBC12345 | FBC6727 | 4,000 | 1/2/2013 |3
Notice how the ranking works based off of the parent Account ID. The account ID DJE523 has the single greatest TCV but it's ranked second b/c the grouped value of parent account ID ParentABC12345 is greater. So there would be a ranking of 35 parent account IDs but in that ranking their could be say 100+ lines of actual data.
Any thoughts?

Answer 1

Always nice to follow up. The "parent rank" is added as an INNER JOIN.
Edit: As correctly mentioned by Dan Bracuk, my first answer was not correct. I altered the query to meet the correct conditions. I also applied the timespan to the Parent Account's.
DECLARE #minimumValue decimal(20,2) = 10000
DECLARE #numberOfAccounts int = 35
DECLARE #from datetime = '1/1/2013'
DECLARE #till datetime = DATEADD(MONTH, 1, #from)
SELECT
[sub].[Sales Rep],
[sub].[Rank],
[sub].[Account ID],
[sub].[Total Contract Value],
[sub].[Parent Account ID],
[sub].[Total],
[sub].[ParentRank]
FROM
(
SELECT
[s].[Sales Rep],
[s].[Account ID],
[s].[Total Contract Value],
DENSE_RANK() OVER (PARTITION BY [s].[Sales Rep] ORDER BY [s].[Total Contract Value] DESC) AS [Rank],
[p].[Parent Account ID],
[p].[Total],
[p].[ParentRank]
FROM [Sales] [s]
INNER JOIN
(
SELECT
[Parent Account ID],
SUM([Total Contract Value]) AS [Total],
RANK() OVER(ORDER BY SUM([Total Contract Value]) DESC) AS [ParentRank]
FROM [Sales]
WHERE[Date] > #from AND [Date] < #till
GROUP BY [Parent Account ID]
HAVING SUM([Total Contract Value]) > #minimumValue
) AS [p] ON [s].[Parent Account ID] = [p].[Parent Account ID]
WHERE [Date] > #from AND [Date] < #till
) AS [sub]
WHERE [sub].[Rank] <= #numberOfAccounts
ORDER BY
[Sales Rep] ASC,
[ParentRank] ASC,
[Rank] ASC
And here is a new Fiddle.

Answer 2

I think this will do it for you, if you're using SQL Server:
Select top 35
SalesRep,
ParentAccountId,
sum(TotalContractValue) from Table
group by SalesRep, ParentAccountId
order by sum(TotalContractValue) desc