Validation rules

Validation, as any other constraint, is a source of assumptions in the application code.

Not only it provides the data quality check at some certain point, the whole logic of processing that piece of data afterwards assumes the validation rule was enforced.

I.e. if you cannot create a customer account without specifying the postal code, you should not be surprised to see SQL statements which will break as soon as we remove this check. The primitive example would be the following SQL query:

SELECT c.FirstName + ' ' + c.LastName + ', ' + c.PostalCode FROM dbo.Customer c WHERE...

So relaxing the validation rule is not as simple as removing the condition check and an error message.
You have to trace the usage of the data and make sure none of the functionality is broken, neither from the technical, nor from the logical point of view.

Say, if you now allow the postal code to be NULL, then firstly, the previous select will now yield NULL in case we have a customer w/o the postal code.

Secondly, the reason might be not that we are not going to send customers their invoices anymore. This might be because now we have a default postal code somewhere else, which should be used when the customer's postal code is not set. I am making this up, but you've got the idea.