In this paper the links from measures of consumption to welfare and well-being are discussed drawing on a study of changes in consumption in Ghana in the 1990s. It is argued that household consumption can act as an opportunity measure of welfare. The widespread distrust of such a welfare measure may be explicable, at least in part, by how misleading it can be if averages of welfare change are presented. In Ghana it has been argued that poverty fell based on a consumption measure of welfare. In this paper it is shown that the welfare measure can be presented informatively in terms of types of households. Doing this for Ghana over the period of the 1990s reveals that while on average per capita consumption rose across all percentiles of the distribution this was not true for farmers who are on average the poorest. We also show how it is possible to present results for household types where certain characteristics remain unchanged. We focus on education and household size. The falls in welfare for farmers nearly doubles if we control for education and household size while the overall rise in consumption per capita is reduced from 11 to 3 per cent per decade. Such simple empirical analysis shows how consumption can be used as an opportunity welfare measure and that a decrease in poverty does not imply at all that most people in the economy had greater opportunities.