69

I'm creating an SQL statement that will return a month by month summary on sales.

The summary will list some simple columns for the date, total number of sales and the total value of sales.

However, in addition to these columns, i'd like to include 3 more that will list the months best customer by amount spent. For these columns, I need some kind of inline subquery that can return their ID, Name and the Amount they spent.

My current effort uses an inline SELECT statement, however, from my knowledge on how to implement these, you can only return one column and row per in-line statement.

To get around this with my scenario, I can of course create 3 separate in-line statements, however, besides this seeming impractical, it increases the query time more that necessary.

SELECT  
    DATE_FORMAT(OrderDate,'%M %Y') AS OrderMonth,
    COUNT(OrderID) AS TotalOrders, 
    SUM(OrderTotal) AS TotalAmount, 

    (SELECT SUM(OrderTotal) FROM Orders WHERE DATE_FORMAT(OrderDate,'%M %Y') = OrderMonth GROUP BY OrderCustomerFK ORDER BY SUM(OrderTotal) DESC LIMIT 1) AS TotalCustomerAmount,

    (SELECT OrderCustomerFK FROM Orders WHERE DATE_FORMAT(OrderDate,'%M %Y') = OrderMonth GROUP BY OrderCustomerFK ORDER BY SUM(OrderTotal) DESC LIMIT 1) AS CustomerID,

    (SELECT CustomerName FROM Orders INNER JOIN Customers ON OrderCustomerFK = CustomerID WHERE DATE_FORMAT(OrderDate,'%M %Y') = OrderMonth GROUP BY OrderCustomerFK ORDER BY SUM(OrderTotal) DESC LIMIT 1) AS CustomerName

FROM Orders     
GROUP BY DATE_FORMAT(OrderDate,'%m%y')
ORDER BY DATE_FORMAT(OrderDate,'%y%m') DESC

How can i better structure this query?


FULL ANSWER

After some tweaking of Dave Barkers solution, I have a final version for anyone in the future looking for help.

The solution by Dave Barker worked perfectly with the customer details, however, it made the simpler Total Sales and Total Sale Amount columns get some crazy figures.

SELECT  
        Y.OrderMonth,   Y.TotalOrders,  Y.TotalAmount,
        Z.OrdCustFK,  Z.CustCompany,    Z.CustOrdTotal, Z.CustSalesTotal   


 FROM 
        (SELECT
            OrdDate,
            DATE_FORMAT(OrdDate,'%M %Y') AS OrderMonth, 
            COUNT(OrderID) AS TotalOrders, 
            SUM(OrdGrandTotal) AS TotalAmount
            FROM Orders
            WHERE OrdConfirmed = 1    
            GROUP BY DATE_FORMAT(OrdDate,'%m%y') 
            ORDER BY DATE_FORMAT(OrdDate,'%Y%m') DESC)
    Y INNER JOIN 
        (SELECT 
            DATE_FORMAT(OrdDate,'%M %Y') AS CustMonth, 
            OrdCustFK, 
            CustCompany, 
            COUNT(OrderID) AS CustOrdTotal,
            SUM(OrdGrandTotal) AS CustSalesTotal 
        FROM Orders INNER JOIN CustomerDetails ON OrdCustFK = CustomerID
        WHERE OrdConfirmed = 1
        GROUP BY DATE_FORMAT(OrdDate,'%m%y'), OrdCustFK 
        ORDER BY SUM(OrdGrandTotal) DESC) 
    Z ON Z.CustMonth = Y.OrderMonth

GROUP BY DATE_FORMAT(OrdDate,'%Y%m')
ORDER BY DATE_FORMAT(OrdDate,'%Y%m') DESC
2
  • 2
    A final note on the full solution is that it reduced my original query time from approx 4 secs down to 0.08 secs Commented Nov 19, 2009 at 6:13
  • 1
    The full solution should be in a separate answer only. Commented Aug 28, 2016 at 2:28

3 Answers 3

36

Move the inline SQL to be a inner join query. So you'd have something like...

SELECT  DATE_FORMAT(OrderDate,'%M %Y') AS OrderMonth, COUNT(OrderID) AS TotalOrders, SUM(OrderTotal) AS TotalAmount,  Z.OrderCustomerFK, Z.CustomerName, z.OrderTotal as CustomerTotal   
  FROM Orders     
  INNER JOIN (SELECT DATE_FORMAT(OrderDate,'%M %Y') as Mon, OrderCustomerFK, CustomerName, SUM(OrderTotal) as OrderTotal 
                FROM Orders 
               GROUP BY  DATE_FORMAT(OrderDate,'%M %Y'), OrderCustomerFK, CustomerName ORDER BY SUM(OrderTotal) DESC LIMIT 1) Z
          ON Z.Mon = DATE_FORMAT(OrderDate,'%M %Y')
    GROUP BY DATE_FORMAT(OrderDate,'%m%y'), Z.OrderCustomerFK, Z.CustomerName
    ORDER BY DATE_FORMAT(OrderDate,'%y%m') DESC
Sign up to request clarification or add additional context in comments.

Comments

12

You can also do something like:

SELECT 
    a.`y`,
    ( SELECT @c:=NULL ) AS `temp`,
    ( SELECT @d:=NULL ) AS `temp`,
    ( SELECT 
          CONCAT(@c:=b.`c`, @d:=b.`d`) 
      FROM `b`
      ORDER BY b.`uid` 
      LIMIT 1 ) AS `temp`,
    @c as c,
    @d as d
 FROM `a`

3 Comments

Thank you. For those wondering, I understand the idea is to do one inline select CONCATenating all you need, then extract separate columns from this, using variables. I think in some cases this is more efficient because subqueries are not always well optimized...
Wicked! Nope @phil_w, the idea is a fake, not used concat to exploit it of the variable setting.
@Julian the first assignments to null are not necessary, I think. Also, using case ... when instead of concat may be better for large strings.
10

Give this a shot:

  SELECT CONCAT(o.order_month, ' ', o.order_year),
         o.total_orders,
         o.total_amount,
         x.sum_order_total,
         x.ordercustomerfk,
         x.customername
    FROM (SELECT MONTH(t.orderdate) AS order_month,
                 YEAR(t.orderdate) AS order_year
                 COUNT(t.orderid) AS total_orders, 
                 SUM(t.ordertotal) AS total_amount
            FROM ORDERS t
        GROUP BY MONTH(t.orderdate), YEAR(t.orderdate)) o
    JOIN (SELECT MONTH(t.orderdate) AS ordermonth,
                 YEAR(t.orderdate) AS orderyear
                 SUM(t.ordertotal) 'sum_order_total',
                 t.ordercustomerfk,
                 c.customername
            FROM ORDERS t
            JOIN CUSTOMERS c ON c.customerid = o.ordercustomerfk
        GROUP BY t.ordercustomerfk, MONTH(t.orderdate), YEAR(t.orderdate)) x ON x.order_month = o.order_month
                                                                            AND x.order_year = o.order_year
ORDER BY o.order_year DESC, o.order_month DESC

1 Comment

Sorry, didn't notice your answer until I posted a revised version of Dave Barkers solution. However, it looks along the same lines as what finally worked for me so thanks.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.