Actuals and variance
Entering actuals closes the loop: it measures how accurate the forecast was and feeds the accuracy metrics that alerting uses.
Pages: Forecasting -> Actuals (/forecasting/actuals), Forecasting -> Import Data (/forecasting/import)
Step 1: Open the actuals page
- In the left sidebar, click Forecasting -> Actuals.
- The page defaults to your default location and its most recent forecast week. Change either dropdown to move around.
If your POS data flows into the warehouse, sales actuals may already be present. Enter actuals manually only for the days that are missing.
Step 2: Enter actuals
-
Click Enter Actuals. A drawer opens.
-
Fill in the fields:
Field Notes Location Pre-selected Date The business day the numbers belong to Net Sales ($) Required Guest Count Optional Check Count Optional Gross Sales ($) Optional -
Click + Add Another Day to keep adding days in one batch.
-
Click Save X Actuals.
Step 3: Read the variance table
The table compares forecast against actual for the selected week:
| Row | Compares |
|---|---|
| Sales | Forecast sales against actual sales |
| Labor | Forecast labor cost against actual labor cost |
| Labor hours | Forecast hours against actual hours |
Cells are color coded: green where the actual came in above the forecast for sales, red where it fell short. A variance within roughly plus or minus five percent is a healthy weekly result.
Import in bulk
Use Forecasting -> Import Data to load history or a full week of actuals at once.
- Go to Forecasting -> Import Data.
- Choose what you are importing and the location.
- Upload the CSV and map the columns to the expected fields (date, net sales, guest count, and so on).
- Review the preview, then confirm the import.
Importing at least a year of daily history is what makes the Smart Forecast YoY model useful for a new location.
What actuals unlock
- Variance analysis on the forecasting dashboard.
- Rolling forecast accuracy (MAPE and bias) used by the forecast-quality alert rules, see Notifications overview.
- Better projections next week, because most models read from the same actuals.
Troubleshooting
| Symptom | Fix |
|---|---|
| Variance table is empty | No published forecast for the selected week |
| Actual saved to the wrong day | Edit the row and correct the date; the table recalculates |
| Accuracy alerts never fire | Actuals are missing for most days; the rules need a populated history |