Sales Forecast Accuracy Report
Generated on 2026-07-22 01:47
Formulas used
Each formula describes a different aspect of forecast quality.
Forecast error
Positive values indicate under-forecasting. Negative values indicate over-forecasting.
Absolute percentage error
APE measures each period's miss relative to its actual value.
MAPE
MAPE is intuitive, but zero actual values require special handling.
WAPE
WAPE emphasizes high-volume periods and handles mixed sales scales well.
RMSE
RMSE gives greater influence to large forecast misses.
Tracking signal
Large absolute signals can indicate persistent forecasting bias.
sMAPE
sMAPE treats actual and forecast values more symmetrically.
MASE
A result below one means the model beats naive persistence.
Forecast Value Added
Positive FVA means the new stage improved forecast accuracy.
Prediction interval coverage
Coverage should be interpreted beside interval width and business risk.
How to use this calculator
A practical workflow for reliable and repeatable forecast reviews.
- Choose the currency, sales unit, headline metric, zero-value rule, and baseline method.
- Enter one row for each comparable forecast period. Supply forecast and actual sales.
- Add product, region, channel, model, bounds, and forecast stages when available.
- Select a scenario adjustment only when testing an alternative planning assumption.
- Set thresholds that match your company’s acceptable planning performance.
- Calculate results, inspect bias and large misses, then compare against the baseline.
- Use FVA results to identify adjustments that improve or reduce forecast quality.
- Export the detailed results and retain assumptions with the planning report.
Understanding sales forecast accuracy
Use several measures together rather than relying on one score.
Why forecast accuracy matters
Reliable forecasts support inventory, staffing, cash planning, and purchasing decisions. Poor forecasts can create shortages, excess stock, missed targets, and unnecessary spending. Accuracy measurement turns forecasting into a controlled business process.
A single percentage rarely explains every planning problem. MAPE is easy to communicate, but it becomes unstable around zero actual values. WAPE is often more useful for portfolios because large periods receive appropriate influence.
Read error and bias together
Absolute errors measure the size of each miss. Signed errors reveal its direction. A model can show acceptable average accuracy while consistently forecasting too high or too low.
Forecast bias matters because repeated direction errors can distort operational decisions. Persistent under-forecasting may cause stockouts or resource shortages. Persistent over-forecasting may lock cash into unnecessary capacity.
Use a baseline before trusting complexity
Every forecasting method should beat a simple reference forecast. Previous-period sales, a moving average, or a seasonal naive estimate can provide that reference. A complex model that fails this test may not justify its maintenance cost.
MASE and the baseline comparison section make this review easier. A MASE result below one indicates that the submitted forecast beat a simple previous-period reference. Baseline improvement expresses the gain as a percentage.
Evaluate human adjustments
Sales teams and managers often adjust statistical forecasts. Some adjustments add customer knowledge that the model lacks. Others introduce optimism, conservatism, or pressure from sales targets.
Forecast Value Added analysis measures these stages separately. Positive FVA means an adjustment reduced MAE. Negative FVA means the intervention made the forecast worse.
Improve the review cycle
Track errors by product, region, channel, customer segment, and forecast horizon. Investigate the largest contributors rather than treating every miss equally. Separate one-time disruptions from repeated process problems.
Review accuracy after actual sales become stable. Keep definitions consistent between forecast and actual values. Document exclusions, scenario changes, promotions, cancellations, currency changes, and unusual events.
Frequently asked questions
Common interpretation and implementation questions.
Which accuracy metric should I use?
Use WAPE for mixed sales volumes, MAPE for simple communication when actuals are nonzero, RMSE when large misses deserve extra attention, and MASE for comparisons against a naive baseline.
Why can forecast accuracy become negative?
The common formula 100 minus APE becomes negative when error exceeds actual sales. Enable clamping when reports must remain between zero and one hundred percent.
How should zero actual sales be handled?
Exclude those periods from MAPE, use WAPE or sMAPE, or replace zero with a documented epsilon. The chosen method should remain consistent across reporting periods.
What tracking signal is acceptable?
Many teams review signals outside approximately minus four to plus four. The correct boundary depends on business volatility, period length, and forecast governance.
Can the calculator evaluate units instead of revenue?
Yes. Enter quantity, orders, customers, profit, or another consistent measure. Change the measurement label so reports describe the selected unit correctly.
Does the calculator save data to a server?
No database is included. Local save uses your browser storage. Submitted data is processed by the current PHP request and is not retained automatically.
How do I create a PDF report?
Use Print or PDF, then choose the browser's Save as PDF destination. The page includes print-specific formatting for cleaner reports.
Can this replace a demand planning system?
No. It supports analysis and review. It does not replace controlled source data, approvals, forecast ownership, or integrated planning workflows.