Crash cost estimates are essential for evaluating road safety management policies and assessing the economic benefits of safety improvements. Existing studies often rely on aggregated crash data, assuming an even distribution of incidents, which overlooks significant spatial variations influenced by road characteristics and traffic conditions. This research presents a methodological framework for link-based crash cost analysis that considers both internal and external costs, enabling detailed quantification at a localized level. By employing safety performance functions and ordered probit models, we estimate on-road crash rates by crash type and injury severity, distinguishing between internal costs borne by individuals involved in crashes and external costs that impact victims, insurers, and government agencies. This framework is applied to the Minneapolis-St. Paul metropolitan area for a proof-of-concept. Our findings reveal that the costs incurred by drivers are higher than those imposed on others, and that highways are generally safer than surface streets. However, these crash costs are too low compared to the value of travel time to significantly influence route choices, even when drivers are aware of these costs. To enhance effective decision-making, related policies should consider offering incentives for safe driving practices. Future research on the practical applications of this framework is encouraged to maintain a dynamic dataset that reflects ongoing changes in road safety conditions.