Corporate social performance (CSP) has become a widely applied concept, discussed in most large firms’ corporate reports and the academic literature alike. Unfortunately, CSP has largely been employed as a way of demonstrating corporate social responsibility (CSR) in practice, or to justify the business case for CSR in academia by relating some measure of CSP to some measure of financial performance. In this article, we discuss multiple shortcomings to these approaches. We argue that (1) CSR activities need to be managed and measured as projects and aggregated to the business or corporate level using a project portfolio; (2) appropriate measures need to be identified that move away from reporting the firm’s activities toward quantifying actual social outcomes achieved; and (3) given the types of projects prevalent in CSR, statistical evaluation methods common in other fields (ideally, pre-test post-test control group designs, such as used in medicine or propensity score matching for ongoing or past projects) should be employed to properly measure outcomes. We make a first, albeit imperfect, attempt at using such an approach with data collected on behalf of the Patrimonio Hoy project, a well-publicized CSR initiative carried out by Cemex in Mexico. We show that the results from this data reinforce concerns voiced earlier in this article.