What makes a new business start-up successful?

This paper seeks a good measure of new business performance, and then explains this measure by various dimensions of business strategy. Three criteria are used to create a one dimensional ordinal ranking of high, medium and low performance for new business starts: employment growth; return on capital employed; and labour productivity. It is shown that statistical cluster analysis provides a convincing separation of a sample of new business starts into high, medium and low performance categories, using a minimum distance criterion for clustering. An ordinal logit model (with selection) is then used to explain this performance ranking. The results indicate that many widely discussed features of small business strategy have little, or even negative, impact on performance. Of the numerous aims that owner managers may adopt (survival, growth etc.), only one appears to have a major impact on performance; the pursuit of the highest rate of return on investment. Many entrepreneurial perceptions of their own capabilities appear false or unimportant, with the exception of organisational features and systems.