T-SQL AVG of multiple columns in a row - sql-server

Question

I'm trying to select the average sales per person per territory out of the AdventureWorks database.
Since this is aggregating multiple columns in a row instead of multiple rows in a column, it seems like I'd need a sub-query, temp table, maybe a CTE, but I'm not sure how to identify which direction to take or how to write it.
Desired result:
| SalesTerritory | SalesPeople | 2011 | 2012 | 2013 | 2014 | AvgSales
+----------------+-------------+--------+---------+---------+---------+----------
| Australia | 1 | NULL | NULL | 184105 | 1237705 | [avg]
| Canada | 2 | 115360 | 3426082 | 2568323 | etc... | [avg]
Code:
SELECT
pvt.SalesTerritory,
COUNT(pvt.SalesPersonID) AS SalesPeople,
SUM(pvt.[2011]),
SUM(pvt.[2012]),
SUM(pvt.[2013]),
SUM(pvt.[2014])
--What's the best way to AVG the sales by year by sales person for each territory here?
FROM
(SELECT
st.[Name] AS [SalesTerritory],
soh.[SalesPersonID],
soh.[SubTotal],
YEAR(DATEADD(m, 6, soh.[OrderDate])) AS [FiscalYear]
FROM
[Sales].[SalesPerson] sp
INNER JOIN
[Sales].[SalesOrderHeader] soh ON sp.[BusinessEntityID] = soh.[SalesPersonID]
INNER JOIN
[Sales].[SalesTerritory] st ON sp.[TerritoryID] = st.[TerritoryID]
INNER JOIN
[HumanResources].[Employee] e ON soh.[SalesPersonID] = e.[BusinessEntityID]
INNER JOIN
[Person].[Person] p ON p.[BusinessEntityID] = sp.[BusinessEntityID]) AS soh
PIVOT
(SUM([SubTotal]) FOR [FiscalYear] IN ([2011], [2012], [2013], [2014])) AS pvt
GROUP BY
pvt.SalesTerritory

Answer 1

You have several options:
1) Use cross apply. Query would look like:
select
*
from
(
--put your query here
) t
cross apply (select AvgSales = avg(v) from (values ([2011]), ([2012]), ([2013]), ([2014])) q(v)) q
2) Count average by yourself
SELECT
pvt.SalesTerritory,
COUNT(pvt.SalesPersonID) AS SalesPeople,
SUM(pvt.[2011]),
SUM(pvt.[2012]),
SUM(pvt.[2013]),
SUM(pvt.[2014]),
ISNULL(SUM(pvt.[2011]), 0) + ISNULL(SUM(pvt.[2012]), 0)
+ ISNULL(SUM(pvt.[2013]), 0) + ISNULL(SUM(pvt.[2014]), 0)
/ CASE WHEN SUM(pvt.[2011]) > 0 THEN 1 ELSE 0 END
+ CASE WHEN SUM(pvt.[2012]) > 0 THEN 1 ELSE 0 END
+ CASE WHEN SUM(pvt.[2013]) > 0 THEN 1 ELSE 0 END
+ CASE WHEN SUM(pvt.[2014]) > 0 THEN 1 ELSE 0 END
FROM
...
GROUP BY
pvt.SalesTerritory

Answer 2

As I understand your question, you need the average of the yearly sales per salesperson, for each territory. Uzi's answer provides the average of the yearly sales for all salespersons together, for each territory. You could divide that result by the number of salespersons, or use a query like this:
SELECT pvt.SalesTerritory, COUNT(pvt.SalesPersonID) AS SalesPeople,
SUM(pvt.[2011]) AS [2011], SUM(pvt.[2012]) AS [2012],
SUM(pvt.[2013]) AS [2013], SUM(pvt.[2014]) AS [2014],
AVG(pvt.AvgSubTotal) AS AvgSubTotal
FROM (
SELECT y.SalesTerritory, y.SalesPersonID, y.FiscalYear, y.SubTotal,
AVG(y.SubTotal) OVER (PARTITION BY y.SalesTerritory) AS AvgSubTotal
FROM (
SELECT x.SalesTerritory, x.SalesPersonID, x.FiscalYear, SUM(x.SubTotal) AS SubTotal
FROM (
SELECT st.Name AS SalesTerritory, soh.SalesPersonID, soh.SubTotal,
YEAR(DATEADD(m, 6, soh.OrderDate)) AS FiscalYear
FROM Sales.SalesPerson sp
INNER JOIN Sales.SalesOrderHeader soh ON sp.BusinessEntityID = soh.SalesPersonID
INNER JOIN Sales.SalesTerritory st ON sp.TerritoryID = st.TerritoryID
INNER JOIN HumanResources.Employee e ON soh.SalesPersonID = e.BusinessEntityID
INNER JOIN Person.Person p ON p.BusinessEntityID = sp.BusinessEntityID
) x
GROUP BY x.SalesTerritory, x.SalesPersonID, x.FiscalYear
) y
) AS soh
PIVOT (SUM(SubTotal) FOR FiscalYear IN ([2011], [2012], [2013], [2014])) AS pvt
GROUP BY pvt.SalesTerritory;
The result is different for the Northwest territory, but I'm not sure which result you want.