I have a calendar table (c) with a set of 3 month dates: I have a Stats table (s) with views of each product on the website. I need to count the views per product per month whether there are views or not. I have followed many SO answers (SQL - Group By with Left Join) and I cannot see

SQL GROUP BY with LEFT JOIN MS SQL Server

I have a calendar table (c) with a set of 3 month dates:

2021-06-01
2021-07-01
2021-08-01

I have a Stats table (s) with views of each product on the website.

Prod1 | 2021-06-01
Prod1 | 2021-06-01
Prod1 | 2021-08-01
Prod2 | 2021-07-01
Prod2 | 2021-08-01
Prod2 | 2021-08-01

I need to count the views per product per month whether there are views or not.

I have followed many SO answers (SQL – Group By with Left Join) and I cannot see a problem with the code below.

DECLARE @Start date
SET @Start=DATEADD(month, DATEDIFF(month, 0,DATEADD(month, -3, getdate())), 0)

SELECT 
s.ProductID, 
c.themonth,
ISNULL(Count(s.ProductID),0) As Total_Views

FROM 
#calendar c

LEFT JOIN
(
SELECT ProductID,FirstDayOfMonth FROM Stats WHERE FirstDayofMonth >= @Start
) s
ON c.themonth = s.FirstDayOfMonth

GROUP BY 
c.themonth,s.ProductID

ORDER BY s.ProductID,c.themonth

I get results only for ProductIDs that have views in a particular month, instead of a row for each ProductID and each Month whether there are views or not.

With the data above, the result I want is:

Prod1 | 2021-06-01 | 2
Prod1 | 2021-07-01 | 0
Prod1 | 2021-08-01 | 1
Prod2 | 2021-06-01 | 0
Prod2 | 2021-07-01 | 1
Prod2 | 2021-08-01 | 2

Answer

Use a cross join to generate the rows and then a left join to bring in the data:

select c.themonth, p.productid,
       count(s.productid) as sales_in_month
from #calendar c cross join
     (select distinct productid from stats) p left join
     stats s
     on s.productid = p.productid and
        s.firstdayofmonth = c.themonth
group by c.themonth, p.productid;

Advertisement

Answer