SQL Get 12 weeks as columns - sql-server

Question

i need help on the following.
I have the basic query below:
select count(transactions)
from tx
where customer = 'AA'
This gives me a count of all transactions for the relevant client.
What I want is a query that gives me the same output but broken down into the LATEST last 12 weeks (Monday-Sunday is one full week). These values should be presented as 12 columns with the header of each column presented as the last date of the week (ie Sunday's date).
Furthermore the total transactions are split into status- failed and success. I would like the rows of the transactions to be failed and success so the final table would look like this:
25/03/2018 (week 1)| 01/04/2018| ........ |17/06/2018 << (week 12)
Success 100 | 200 | ........ | 150
Failed 3 | 4 | ........ | 6
Any ideas how this can be done?
Thanks you in advance

Answer 1

Returning pivoted data is usually a lot more hassle than it is worth and you should just leave this up to your presentation layer, which will handle your dynamic columns with much more grace. Regardless of the presentation layer you are using (SSRS, Excel, Power BI, etc), you will get the most flexibility by providing it a standard set of unpivoted data:
declare #t table (id int, TransactionDate date, Outcome varchar(8));
insert into #t values
(1,getdate()-1,'Success')
,(2,getdate()-2,'Success')
,(3,getdate()-2,'Success')
,(4,getdate()-3,'Success')
,(5,getdate()-6,'Failed')
,(6,getdate()-6,'Success')
,(7,getdate()-7,'Success')
,(8,getdate()-8,'Success')
,(9,getdate()-8,'Success')
,(10,getdate()-10,'Success')
,(11,getdate()-10,'Failed')
,(12,getdate()-11,'Success')
,(13,getdate()-13,'Success')
;
with w(ws) as(select dateadd(week, datediff(week,0,getdate())-w, 0) -- Monday at the start of the week, minus w.w weeks for all 12
from (values(0),(1),(2),(3),(4),(5),(6),(7),(8),(9),(10),(11)) as w(w)
)
,d(ws,d) as(select w.ws
,dateadd(day,d.d,w.ws) as d -- Each day that makes up each week for equijoin to Transactions table
from w
cross join (values(0),(1),(2),(3),(4),(5),(6)) as d(d)
)
select d.ws as WeekStart
,t.Outcome
,count(t.TransactionDate) as Transactions
from d
left join #t as t
on d.d = t.TransactionDate
group by d.ws
,t.Outcome
order by d.ws
,t.Outcome;
Output:
+-------------------------+---------+--------------+
| WeekStart | Outcome | Transactions |
+-------------------------+---------+--------------+
| 2018-04-09 00:00:00.000 | NULL | 0 |
| 2018-04-16 00:00:00.000 | NULL | 0 |
| 2018-04-23 00:00:00.000 | NULL | 0 |
| 2018-04-30 00:00:00.000 | NULL | 0 |
| 2018-05-07 00:00:00.000 | NULL | 0 |
| 2018-05-14 00:00:00.000 | NULL | 0 |
| 2018-05-21 00:00:00.000 | NULL | 0 |
| 2018-05-28 00:00:00.000 | NULL | 0 |
| 2018-06-04 00:00:00.000 | NULL | 0 |
| 2018-06-11 00:00:00.000 | NULL | 0 |
| 2018-06-11 00:00:00.000 | Success | 2 |
| 2018-06-18 00:00:00.000 | NULL | 0 |
| 2018-06-18 00:00:00.000 | Failed | 2 |
| 2018-06-18 00:00:00.000 | Success | 5 |
| 2018-06-25 00:00:00.000 | NULL | 0 |
| 2018-06-25 00:00:00.000 | Success | 4 |
+-------------------------+---------+--------------+