| Abstract: | In trying to explain the difficulties of implementing evaluation models it can be useful to study how evaluations of product development proposals are carried out in practice when they are not influenced by normative models. In this article some results are reported from an in-depth study of how project proposals were evaluated in two Swedish companies operating in the same industry. One was among the most innovative in the industry, its product development activities had led it into a series of new products and had resulted in a very high profitability and an extremely high rate of growth. The product development of the other company had resulted much more in variations in existing products than in major innovations. Growth was very slow and profitability was approaching a critically low level. The way project proposals were evaluated in the companies proved to differ substantially. In one company the dominating evaluation mode was one where the task is considered to be to estimate what effect an acceptance of the project proposal will have on overall goals, such as profitability and growth. This is called the rationalistic approach. In the other company an evaluation mode with the opposite characteristics was used. The evaluators consider a few conspicuous or at least easily visible characteristics of the project proposal. On the basis of these a picture of the project is created which is perceived as good or bad. This picture is the basis for the decision. This is called the impressionistic approach. In the case reported here the rationalistic mode was used by the least successful and least innovative company while the impressionistic mode was used by the most successful and most innovative company. The paper presents arguments for interpreting the relationship between evaluation mode and performance not as coincidental but as causal. It is argued that even if the impressionistic mode may seem irrational from a traditional point of view, it is superior to the rationalistic one in some important respects. |
