At some point the penny will drop for you and you will realise that you can generate data structures in Perl to match whatever you are trying to achieve. But first you need to figure out what you want to achieve.

Indeed, in Perl it is generally trivial to match a structure to the form of your data, although it is not always so easy to manage the structure you've created. Consider for example you have a file containing groups of three lines where the first line is a header, the second line is a list of objects and the third line a mapping between objects and names. You could end up with a structure that looks like:

my @records = (
    ["Record 1", [qw(obj1 obj2 obj3)], {obj1 => ['Sam'], obj3 => ['Bob
+']}],
    [
        "Record 2", [qw(obj1 obj5 obj6)],
        {obj5 => ['Lew', 'Fred'], obj6 => ['Bob']}
    ],
);
[download]

So, what are you going to call that puppy? When your structures start getting that interesting it is time to wrap it up with some code, most likely by turning the elements of @records into Record objects and provide appropriate accessor methods for the various parts of the record.

Actually, walkingthecow, I find your code easier to read and understand.

Your comment implies that my code and the OP's code are functionally equivalent. They are not.

I urge you to expend a modicum of effort and do the following simple tasks:

• Download and run the OP's code. You will discover that it does not do what the OP wants it to do. It only stores one email address per category (sales, etc.). This is the reason the OP has asked for help.
• Perl's motto is "There's More Than One Way To Do It". Please show us another way that is easy to read and understand, with the constraint that it does what the OP wants it to do.