Fornell et al. (1996) have proposed and applied the model of the American Customer Satisfaction Index (ACSI). This article reports a research undertaken to empirically test the adjustment of that model with the clients of the Brazilian automobile industry using structural equation modeling. In a survey, using the internal record of 7 dealerships in Belo Horizonte, MG, data were collected from 304 clients, comprising owners of cars manufactured by Fiat, Ford, General Motors and Volkswagen. Taking the model as a whole and in the limits of the sample generated, everything seems to indicate that it suffers from lack of adjustment, which brings two fundamental implications. On the one hand, there is the perspective stating that statistical tests may be a rather poor tool to choose a model, and as such statistical rejection would not prevent practical. On the other hand, a more purist approach would lead to the rejection of the ACSI model, as it is now designed, in the context of the industry examined. No matter the option chosen, it would be great to count with a brazilian index of customer satisfaction in the future, which could provide great potential benefits to clients and suppliers.
model; index; customer satisfaction; United States; Brazil
