SQL count and compare two created_at columns - sql

Question

I have got a table as follows,
ID
CreatedAt_1
CreatedAt_2
ABC
2022-06-10 20:28:37
CFR
2022-06-13 10:00:12
2022-06-10 20:28:14
PFR
2022-06-17 12:20:40
XYZ
2022-06-15 11:00:12
2022-06-10 16:45:05
DFL
2022-06-13 15:00:06
FGT
2022-06-20 10:00:20
2022-06-10 13:34:55
I already used this query to count number of rows on specific date for each column separately :
SELECT
(CAST(datetrunc(‘day’, ‘createdAt_1’ + (INTERVAL '1 day'))) AS timestamp) + (INTERVAL '-1 day')) AS ‘new user’,
count(*) AS ‘count’
FROM Table
WHERE time_interval
GROUP BY ‘new user’
And get something like :
Day
Count
2022-06-10
1
2022-06-13
2
2022-06-15
1
I would like to be able to compare both columns and get percentage on specific day count(createdAt_1) / count(createdAt_2) * 100 but i don’t see how to easily do it.

Answer 1

I wasn't able to verify is Metabase SQL supports the SQL standard coalesce() function, but the purpose of that function is to return the first non-null value it encounters amongst the parameters passed into it (left to right). If it is supported I suggest the query below.
SELECT
datetrunc('day', coalesce('createdAt_1', 'createdAt_2')) AS "new user"
, count(*) AS "count"
FROM TABLE
-- WHERE time_interval is true ??
GROUP BY
datetrunc('day', coalesce('createdAt_1', 'createdAt_2'))
coalesce('createdAt_1', 'createdAt_2') would return the value in createdAt_1 if that isn't NULL, but would return the value in createdAt_2 if createdAt_1 is NULL. If both columns are NULL then overall it returns NULL.
typically labels are denoted by double quotes e.g. "new user"
I don't believe you need to add or subtract intervals, or convert to timestamp. The datetrunc() function should be sufficient in itself.
I have not used the "new user" alias in the group by clause as this is my preference
It isn't clear to me what that where clause is achieving.