Data Analytics for Revenue Forecasting
A retail company is implementing a new data analytics platform to improve revenue forecasting. The platform uses historical sales data, seasonality, and external economic indicators to predict monthly revenue. The finance team has been using a simple moving average method, which has an average forecast error of 12%. The new platform's pilot test on the last 12 months of data shows a mean absolute percentage error (MAPE) of 8%. However, the pilot also revealed that the platform's forecasts are highly sensitive to the quality of input data. During the pilot, two months had incomplete sales data due to a system outage, and the platform produced forecasts with errors of 25% and 30% for those months. The cost of the platform is $200,000 per year, and the finance team estimates that a 1% reduction in forecast error could save $50,000 annually through better inventory management and reduced stockouts. The CFO is excited about the potential savings but is concerned about data quality issues.
1. Calculate the expected annual savings from the new platform based on the pilot's average MAPE improvement, ignoring data quality issues. Show your work.
Read the worked answer
Current MAPE = 12%, New MAPE = 8%, Improvement = 4 percentage points. Savings per 1% reduction = $50,000. Expected savings = 4 × $50,000 = $200,000. Net benefit = $200,000 - $200,000 = $0. So, ignoring data quality, the platform breaks even.
2. Considering the data quality issues observed in the pilot, what additional analysis or controls should the finance team implement before deciding to adopt the platform? Explain how you would adjust the expected savings calculation.
Read the worked answer
The finance team should: (1) Assess the frequency and impact of data outages; (2) Implement data quality controls (e.g., validation checks, redundancy); (3) Recalculate expected savings by incorporating the probability of data quality issues. For example, if data outages occur 2 months per year and cause an average error increase of 15 percentage points (from 8% to 23%), the average MAPE over the year would be (10 months × 8% + 2 months × 23%) / 12 = (80 + 46)/12 = 10.5%. Improvement over current 12% is 1.5 percentage points, yielding savings of 1.5 × $50,000 = $75,000, which is less than the $200,000 cost. Thus, the platform would not be justified unless data quality is improved.