SQL : select distinct from two columns and min from the third colum

My data set looks like this

I want to run a SQL query that returns this output

I effectively want to use select distinct on customer and product and find the min purchase date for that distinct combination, I have tried select statements, tried partitions but the output isn't returned as expected.

When I use partitions, the minimum date from the entire column gets picked up as opposed to min within each distinct customer & product combination.

Any help would be appreciated.

with data as 
(
    select 
        *, 
        row_number() over (partition by customer order by purchase-date) as rownum
    from 
        table
)
select * 
from data 
where rownum = 1

This is supposed to return a rownum column and then I run a query with a where clause rownum = 1, but when I do that it picks up the lowest value of purchase-date and not lowest in each

select distinct does not allow calculations such as minimum, but GROUP BY can, while it also creates distinct combinations:

SELECT 
    customer, product, MIN(purchase-date) as purchase-date
FROM t
GROUP BY
    customer, product

Put the columns that you want as the distinct items under the group by clause, and put those plus the MIN or MAX (or SUM COUNT etc.) in the select clause.

To use row_number() to achieve the equivalent, you need to specify all the columns that you need to be "distinct" in the partition by element

SELECT 
     *
FROM (
    SELECT 
         customer, product, purchase-date
       , row_number() over(partition by customer, product 
                           order by purchase-date ASC) as rn
    FROM t
     ) d
WHERE rn = 1

Row numbering (re)commences at 1 for each distinct combination of customer, product, and the specific row that will get row 1 is the row with the earliest date (or it could be equal earliest). Only 1 row for each partition can get the row number 1.

Tests:

CREATE TABLE t (
  customer VARCHAR(50),
  product VARCHAR(50),
  purchase_date DATE
);
 
INSERT INTO t (customer, product, purchase_date)
VALUES
  ('vik', 'office', '2022-04-21'),
  ('vik', 'office2', '2021-02-21'),
  ('vik', 'office', '2021-01-21'),
  ('vik', 'office2', '2023-02-21'),
  ('abc', 'office', '2022-02-01'),
  ('abc', 'office', '2021-02-01');

SELECT 
    customer, product, MIN(purchase_date) as purchase_date
FROM t
GROUP BY
    customer, product

SELECT 
     *
FROM (
    SELECT 
         customer, product, purchase_date
       , row_number() over(partition by customer, product 
                           order by purchase_date ASC) as rn
    FROM t
     ) d
WHERE rn = 1

fiddle