Oracle 11g Release 2

By using db<>fiddle, you agree to license everything you submit by Creative Commons CC0.
create table test (
  SalesId int,
  ProductId int, 
  ProductName varchar2(50),
  SalesDate date
);
begin
  insert into test values(1, 1, 'Football', date'2020-01-07');
  insert into test values(2, 1, 'Football', date'2019-01-08');
  insert into test values(3, 1, 'Football', date'2019-01-08');
  insert into test values(4, 1, 'Football', date'2019-01-08');
  insert into test values(5, 2, 'Racket', date'2020-01-07');
  insert into test values(6, 2, 'Racket', date'2018-01-07');
  insert into test values(7, 2, 'Racket', date'2018-01-07');
  commit;
end;
/
1 rows affected

select pid, salyr, cnt, 
LAG(cnt, 1, 0) OVER (PARTITION BY pid ORDER BY salyr) AS sal_prev,
cnt - LAG(cnt, 1, 0) OVER (PARTITION BY pid ORDER BY salyr) AS sal_diff,
case when LAG(cnt, 1, 0) OVER (PARTITION BY pid ORDER BY salyr) = 0 then 0
else 
round((cnt - LAG(cnt, 1, 0) OVER (PARTITION BY pid ORDER BY salyr))*100
/(LAG(cnt, 1, 0) OVER (PARTITION BY pid ORDER BY salyr)),2) end as perc
from
(select 
pid, salyr, count(1) cnt,
ROW_NUMBER() OVER (PARTITION BY pid ORDER BY salyr ASC ) AS rn
from
(select 
productid as pid, extract(year from salesdate) as salyr
from test)
group by pid, salyr
order by 1, 2)
;


PID
SALYR
CNT
SAL_PREV
SAL_DIFF
PERC


1
2019
3
0
3
0

1
2020
1
3
-2
-66.67

2
2018
2
0
2
0

2
2020
1
2
-1
-50