If we determine that support IE6 will likely not matter, we break it out as an extra line-item cost and give them the option to support IE6 by paying extra. We explain the facts of their situation and let them make the decision.
It's not about making them feel bad. It's just human nature. It's just a more pleasant experience just paying a price and knowing it'll just work instead of having to pay more than the regular price just to make it work.
Like, "for this site and audience, we recommend supporting IE6. But FYI, that does eat this much of our costs, which can't be spent on shiny new features for modern browsers. So maybe for other projects, if you do something that targets a different audience, we'll want to make a different trade-off."
If I were the client, I think I'd appreciate that info. It might affect what types of sites I want to launch, for example.
For involved sites, it just depends what is being done and whether you're trying to replicate the experience or if you're just creating a graceful degradation for IE6 users, which is often "good enough" for a lot of clients.