Get number of times a user has availed a particular offer - sql

Question

I have a table which gives information about when a particular user has used an offer. It has 3 columns
Date: Date at which the offer was used
user_id: Identifier for a particular user
txn_id: Transaction id when a user uses an offer. It is always unique in the table.
The offer is such that a particular user can use it for 5 times.
I want to know at each date the number of users are in which stage of offer usage.
For example
On Day 1 there could be 3 users who have used offer once(redemption_1), 2 users who could have used offer twice (redemption_2).
Now on Day 2 there could be users from day 1(repeat users) as well as users who are coming for offer usage for the first time(new users).
For the new users of day 2 the logic is same as that of day 1 users.(May be 2 new users use the offer for 1 time(redemption_1), 3 new users use it for 3 times(redemption_3))
But for the repeat users now I want to add up to there previous day's usage.
For example
On Day 1, 3 users had used offer once(redemption_1) but on day 2 if they use it one more time then they should be counted in redemption_2.(And not in redemption_1 since they are using it for second time since the offer has started/or there last usage)
In this way I want to go on adding cumulatively the number of time a user has used a offer and the count the number of users who have used offer for 1 time(redemption_1), 2 time(redemption_2) and so on for each date
Table
+------------+---------+------------+
| Date | user_id | txn_id |
+------------+---------+------------+
| 2019-06-04 | 1 | 1ACSA0-ABA |
| 2019-06-04 | 2 | 1BEAA0-CSC |
| 2019-06-04 | 3 | 1AGHF0-CBA |
| 2019-06-04 | 1 | 1AVFA0-GAA |
| 2019-06-05 | 1 | 1BCFA0-AAA |
| 2019-06-05 | 1 | 1AVFB0-GAC |
| 2019-06-05 | 2 | 1AVFA0-GVA |
| 2019-06-05 | 4 | 1AVFA0-GVB |
| 2019-06-05 | 5 | 1AVFA0-BCF |
| 2019-06-06 | 6 | 1AGHF0-CCA |
| 2019-06-06 | 1 | 1BXHF0-CCA |
| 2019-06-06 | 2 | 1AGHF0-CBG |
| 2019-06-06 | 3 | 1AGHF0-CAW |
| 2019-06-06 | 2 | 1AGHF0-CTU |
+------------+---------+------------+
Desired Output
+------------+--------------+--------------+--------------+--------------+--------------+
| Date | redemption_1 | redemption_2 | redemption_3 | redemption_4 | redemption_5 |
+------------+--------------+--------------+--------------+--------------+--------------+
| 2019-06-04 | 2 | 1 | 0 | 0 | 0 |
| 2019-06-05 | 2 | 1 | 0 | 1 | 0 |
| 2019-06-06 | 1 | 1 | 0 | 1 | 1 |
+------------+--------------+--------------+--------------+--------------+--------------+
I will walk you through the rows of output for better understanding
In row one with date 2019-06-04 there are two users who used offer once (2,3) and one user who used offer twice(1)
In row with date 2019-06-05 there are 2 user who used offer once(4,5). Note that they have never used offer before that so they counted for redemption_1.
In the same row there is 1 user who has used offer 2 times (2: Once on 2019-06-04 and then on 2019-06-05) so he is counted for redemption_2
In the same row there is 1 user who has used offer 4 times (1: Twice on 2019-06-04 and then again twice on 2019-06-05) so he is counted for redemption_4
And so on for row with date 2019-06-06
Please let me know for any kind of clarification

Answer 1

Not a paragon of efficiency, but it works.
Test data:
Create Table offer_used(date DateTime, user_id Int, txn_id Varchar(50))
Insert Into dbo.offer_used (date,
user_id,
txn_id)
Values
('2019-06-04', 1, '1ACSA0-ABA'),
('2019-06-04', 2, '1BEAA0-CSC'),
('2019-06-04', 3, '1AGHF0-CBA'),
('2019-06-04', 1, '1AVFA0-GAA'),
('2019-06-05', 1, '1BCFA0-AAA'),
('2019-06-05', 1, '1AVFB0-GAC'),
('2019-06-05', 2, '1AVFA0-GVA'),
('2019-06-05', 4, '1AVFA0-GVB'),
('2019-06-05', 5, '1AVFA0-BCF'),
('2019-06-06', 6, '1AGHF0-CCA'),
('2019-06-06', 1, '1BXHF0-CCA'),
('2019-06-06', 2, '1AGHF0-CBG'),
('2019-06-06', 3, '1AGHF0-CAW'),
('2019-06-06', 2, '1AGHF0-CTU')
Query:
; With
Dates As (Select Distinct date From dbo.offer_used OU),
Users As (Select user_id, FirstTime = Min(date) From dbo.offer_used OU Group By user_id),
UserCounts As (Select
Dates.date,
Users.user_id,
Users.FirstTime,
UsedCount = (Select Count(*) From dbo.offer_used As Used
Where Used.date <= Dates.date
And Used.user_id = Users.user_id)
From
Dates
Cross Join Users)
Select
date = UserCounts.date,
[first time today] = Sum(Case When UserCounts.date = UserCounts.FirstTime
And UserCounts.UsedCount = 1 Then 1 Else 0 End),
[2 times total] = Sum(Case When UserCounts.UsedCount = 2 Then 1 Else 0 End),
[3 times total] = Sum(Case When UserCounts.UsedCount = 3 Then 1 Else 0 End),
[4 times total] = Sum(Case When UserCounts.UsedCount = 4 Then 1 Else 0 End),
[5 times total] = Sum(Case When UserCounts.UsedCount = 5 Then 1 Else 0 End),
[bonus: never] = Sum(Case When UserCounts.UsedCount = 0 Then 1 Else 0 End)
From UserCounts
Group By UserCounts.date
Order By UserCounts.date
Results:
date first time today 2 times total 3 times total 4 times total 5 times total bonus: never
----------- ---------------- ------------- ------------- ------------- ------------- ------------
2019-06-04 2 1 0 0 0 3
2019-06-05 2 1 0 1 0 1
2019-06-06 1 1 0 1 1 0

Answer 2

I think you want conditional aggregation:
select t.date,
sum(case when seqnum = 1 then 1 else 0 end) as redemption_1,
sum(case when seqnum = 2 then 1 else 0 end) as redemption_2,
sum(case when seqnum = 3 then 1 else 0 end) as redemption_3,
sum(case when seqnum = 4 then 1 else 0 end) as redemption_4,
sum(case when seqnum = 5 then 1 else 0 end) as redemption_5
from (select t.*, row_number() over (partition by user_id order by date) as seqnum
from table t
) t
group by t.date
order by t.date