To explain a forecast change, compare two recorded versions for the same scope and then investigate the evidence for the movement. Today’s figures and an update timestamp alone cannot establish what changed or why.
Identify which forecast changed
Record the property, department, dates, measure and report. Distinguish the submitted Forecast or Budget from the rolling Live forecast, PMI Prediction and an imported external forecast. Also check the selected units, currency, segments and filters.
The Live forecast combines actuals for completed days with the outlook for remaining days. As time passes, replacing an expected day with its actual outcome can change a period total even if no user manually edits the remaining forecast.
Compare like-for-like versions
Use dated snapshots, retained exports or other available records of yesterday’s and today’s values. Record when each was captured, the business dates covered and whether the values were saved or only displayed as unsaved edits.
Calculate the difference by date and, where available, by segment or department. Separate changes in volume, rate and revenue. Do not add a parent total to its contributing rows.
Illustrative example: yesterday’s full-month Live forecast was 1,000,000 currency units; today’s is 1,030,000. Matched daily records show that replacing an expected day with actual revenue adds 10,000 and changes to the remaining outlook add 20,000. The net movement is 30,000. This quantifies where the change occurred; it does not yet identify why the future outlook changed.
Check the possible sources of the movement
| Evidence to inspect | What it may help establish |
|---|---|
| New actuals or OTB import | Which source values changed and when they arrived |
| Automatic prediction or linked driver update | Whether an updated input or automated forecast contributed |
| Manual edit or saved pickup | Whether a recorded adjustment affected the relevant dates |
| Forecast mode change or reset | Whether the day changed between manual, automated or imported handling |
| Availability, mapping or other setup change | Whether the inputs or reporting basis changed |
| Plan submission or selected version | Whether the report now uses a different saved comparison |
Use the forecast configuration and any available change records to test these possibilities. Some imported forecasts can replace manual adjustments; behaviour depends on the configured workflow. Check the applicable settings before attributing a change to the robot or a user.
State what the evidence proves
A useful conclusion is: “The change is concentrated on these dates, and the available records show that their imported forecast values changed.” A stronger claim about the cause requires the corresponding source or audit evidence.
If only the latest values are available, explain that the historical difference cannot yet be reconstructed. Request a retained export, snapshot or change record from the responsible owner. Do not fabricate yesterday’s forecast or name an editor from an update time alone.
Decide whether action is needed
Compare the changed outlook with current demand and the property’s assumptions. An automatic change is not necessarily an error, and restoring yesterday’s number is not necessarily the right response.
If a source or configuration error is confirmed, use the authorised correction process and verify the saved values after the next relevant update. Otherwise, agree whether the new outlook should change staffing, purchasing or another operating plan.