This article takes as its background the shift over time in Britain towards models of target setting in education which are based almost entirely on measurable outputs. The original 'headline' targets proposed by the National Council for Educational Technology were of a very different nature, as are those of many other developed nations. The article outlines some alternative models involving targets for both inputs and processes, and the use of targets which are exhortative rather than directly measurable. It is suggested that the main reason these more 'qualitative' targets have disappeared from British policy-making is that they are not considered to be directly measurable. This is contrasted, at least in some accounts, with the more visible output indicators. However, the substantive purpose of this article is show how weak the claim is that we can ever adequately measure even these privileged targets. The article therefore argues that if a policy of target setting is to be pursued, and given other findings from the authors' study in Wales this is far from clear, then more sophisticated models than those in current vogue may be required. The expense involved in setting up such an extension to our already heavily audited society may be unjustified-for it is surely better to spend money on improving lifelong learning even if we cannot measure that improvement definitively, than to have an accurate metric revealing a picture of stagnation or worse.