CustomerId, Location
====================
Microsoft, US
Microsoft, India
Microsoft, Japan
Sun, Japan
Sun, India
Sun, US

Purchase
========
Microsoft, US
Sun, India

Now if I want to target all those Customers in Japan who have not yet made a purchase, I would use the condition table2.country=’Japan’. Otherwise I would get both India and Japan for a customer Microsoft, and US and Japan for Sun.

regards
Anant

Sometimes such results are needed. Lets say you have a customer table and a purchases table, and you want to target all the customers ‘that have not made a purchase yet’ for a special campaign.

This can typically be solved as:

select distinct cust_id from customers
minus
select distinct cust_id from purchases

However, this approach cannot be used for very large tables (say for telecom data) and in those cases the technique outlined in the post can be used.

Guess I should have explained this in the post itself – as the technique and its usage is non-obvious.

Glad you asked!

Thanks,

Brian