[ptrklly]

Interesting! Do you know how they value the negative PR cost? My understanding is that "brand value" is notoriously hard to measure, which made me think it would be difficult to include in the type of cost/benefit comparison you refer to (the $.05 vs $.01 per pound cost).

[reitanqild]

There are some interesting techniques. Off the top of my head there is the shopping basket experiement:

Show shopping lists to the participants. Both lists are equal except one product, the one you want to measure. Ask participant what they can tell you about whoever created the list.

Example: two similar lists, one with standard coffee and one with instant coffee. Result: They found that instant coffee was associated with "lazy" etc (they hoped for "smart" and subsequently adjusted their branding to achieve this.)

(I guess it's just a special case of A/B test)

Edit:

I read about this in school many years ago but at least here's one source:

http://web.utk.edu/~rhovland/PTsandCR.html : Projective Techniques in Consumer Research

(My google query was: instant coffee grocery list experiment)