Having sample data is important to getting assistance. It's still difficult to determine how your data looks. Let's assume it looks like this:
create table table1 (
client varchar(10),
product varchar(10)
);
insert into table1 values
('xxx', 'A'),
('bbb', 'A'),
('bbb', 'A'),
('ccc', 'A'),
('ddd', 'A'),
('ddd', 'A'),
('ddd', 'A'),
('ddd', 'A'),
('tt', 'A'),
('tt', 'A'),
('tt', 'A'),
('tt', 'A'),
('bdad', 'A'),
('bdad', 'A'),
('bdad', 'A'),
('bdad', 'A');
I don't have access to a DB2 database, but this query works for most dbms types. You may need to tweak to fit DB2.
select purchased as most_common_value
from (
select client, count(*) as purchased
from table1
where product = 'A'
group by client
)z
group by purchased
order by count(client) desc
limit 1
Output of query is:
most_common_value
4