I work in market research and this is the classic problem with 'importance' questions. In practice what the survey should be doing is trying to understand the trade-offs being made between different levels of service provision - conjoint analysis is one such technique for doing this, but it's a bit beyond simple questionnaires