You have no idea if a fifth of IE6 users will use your products: for all you know they could all use it.

My advice: support IE6 entirely, launch, iterate, then once you have users (= market) then make the choice to scale back IE6 development.

As a young startup you'll have plenty of reasons why users won't stick - don't make browser support (eg, something you can fix and are aware of) another reason.