In this paper we revisit the relative forecasting improvement from the use of an industry-specific model compared to that of an economy-wide model. We extend the results of Schroder and Yim (2018) who demonstrate that the industry-specific model only improves forecasting for single-segment firms due to the complication of different industries confounding the industry benchmark for multiple segment firms. To explore the issues, we argue that the level of diversification is a more nuanced matter than a simple single-/multiple-segment dichotomy. Our results suggest that industry-specific models can provide more accurate forecasts of future performance and growth, not only for single-segment firms, but also where the level of diversification is relatively low. The results help contribute to the forecasting literature to further explain when the use of different benchmarks are useful in predicting future performance.