In this article, we assess fund performance using data envelopment analysis (DEA). For each inefficient fund, DEA provides a set of role model funds whose best practices may be emulated for performance improvement. We find that the role models of most inefficient funds consist entirely of funds different from their own type. To overcome this situation, we suggest a multi-step DEA procedure. The procedure starts by categorizing funds on a hierarchical basis. We establish the hierarchy based on the frequency of efficient funds that belong to each fund type. Thereafter, a set of role model funds for each inefficient fund is found by pooling the funds in its own category and the funds that belongs to the categories at the lower levels in the hierarchy and applying DEA. This procedure is repeated by augmenting the pool with funds at the next higher level and so on until all the sampled funds are included. At each step, a set of role models is identified. An inefficient fund can thus reach the efficient frontier in stages. Statistical evidence suggests that membership and proportion of risky assets may have a negative association, and the fund size may have a positive association with fund performance.