Sustainability Indicators and Product Diversification: An Exploratory Predictive Assessment

ABSTRACT

This study examines whether carbon footprint, eco‐friendly manufacturing and recycling programmes help predict recorded product‐line count, used as a limited indicator of product diversification. Using the records from a sustainability dataset, we apply exploratory predictive modelling with linear regression, random forest and gradient boosting. We evaluate sustainability‐only, context‐only and combined models against a training‐mean benchmark using a held‐out sample and fivefold cross‐validation. Adding sustainability information improves some predictions beyond price, country, year and market trend but offers little consistent advantage over simply predicting the average product‐line count. The study contributes by showing the difference between improving a model and providing useful predictions of product‐range size. Uncertain corporate provenance and undocumented product‐line definitions prevent broader conclusions about sustainability's effects on strategic diversification or stakeholder value.