Some members of the ubiquitous computing community feel that application-led research needs to make more coherent progress. Their perception is that with few exceptions, such research is neither systematically building on what little new knowledge it has derived so far nor setting specific challenges and benchmarks to guide its progress. In light of these concerns, the 2005 UbiApp workshop's tenet was that ubiquitous computing research could benefit from better metrics for the selection, analysis, and evaluation of applications and common infrastructure. The workshop aimed to identify methodological problems in the way researchers conduct application-led research and to recommend how to address these problems.