Maybe if you ran some kind of candy A/B Test.
Send two different kinds of envelopes to people, and keep track of who got what. Then, send a survey to these people to have them rate the candy's they received. After a few months, you should start to have a general idea on what types of candies are liked more and what types are liked less. Even though you'll never have a taste-consensus, you can get a general idea, and modify what kinds of candy you send and the amounts of the kinds of candies you do send.
Or, another way would be to send all the same candy to everyone and send out a survey. Then the next two weeks send another set of candy and run another survey.
Whatever you do / don't do, getting stats and info is really nice.
You can't, unfortunately, please everyone :( so might as well please a group of people _a lot_ and get a group of really evangelical fans, I think.