The model is accurate enough that, on average across all vehicles, it demonstrably achieves higher returns than any blanket platform assignment. It operates efficiently even with large data volumes and delivers interpretable results.

What matters in practice: the model doesn't just deliver a prediction, but also a confidence score. At high confidence, it routes precisely. At lower confidence, it recommends broader distribution across multiple platforms.