Direct answer
How do students build a useful business forecast?
A useful forecast connects a time period, measurable outcome, decision settings, and assumptions. Students estimate a base case, then define a plausible low and high case by changing one uncertain driver at a time. The range is not a promise; it is a transparent planning boundary.
After the simulation reveals an actual result, students calculate error, identify which assumption contributed most, and update the next period without rewriting the original forecast. This separates honest learning from hindsight and teaches that forecasting is a repeatable decision process, not guessing the winning number.
Forecast menu
Six scenarios with decision-relevant outcomes
| Simulation | Forecast | Uncertain driver | Planning decision | Worksheet |
|---|
| Lemonade stand | Cups sold or profit | Demand and weather | Batch size and price | Open |
|---|
| Restaurant | Customers or revenue | Demand and service time | Staffing and capacity | Open |
|---|
| Motel | Occupied rooms | Demand and price response | Rate and service level | Open |
|---|
| Bakery | Units sold or waste | Morning demand | Production quantity | Open |
|---|
| Coffee shop | Transactions or profit | Traffic and attach rate | Labor and inventory | Open |
|---|
| Childcare | Enrollment or cash | Enrollment and staffing | Capacity planning | Open |
|---|
Forecast one primary outcome and track one guardrail such as waste, waiting, quality, staffing readiness, or cash. Never recommend exceeding a safety, licensing, or capacity constraint to hit a target.
Ready-to-use sequence
50-minute forecast–test–update lesson
Learning goal: create a range forecast from stated assumptions, measure error without hiding misses, and revise a decision when evidence changes.
- Frame the decision — 5 minutes. Choose the scenario, forecast period, one primary outcome, one guardrail, and the decision the forecast will inform.
- Build the driver map — 7 minutes. List three inputs that could move the outcome. Mark each as a chosen setting, model condition, or unknown.
- Set low, base, and high — 8 minutes. Write one value for each case and the assumption that separates it from the others. Check that low ≤ base ≤ high.
- Commit before testing — 3 minutes. Circle the base forecast and timestamp the record. Do not edit the original after viewing results.
- Run and record — 9 minutes. Use the documented settings, complete one period, copy the actual primary and guardrail outcomes, and note any unexpected condition.
- Diagnose error — 8 minutes. Calculate signed and absolute error, check whether the actual fell inside the range, and identify the assumption most in need of revision.
- Issue a rolling update — 10 minutes. Forecast the next period, name exactly what changed, choose the planning action, and state what evidence would trigger another revision.
Printable student record
Forecast, actual, and update sheet
Business and scenario: ______________________________________________
Forecast period: __________________ Decision it informs: __________________
Primary outcome and unit: __________________ Guardrail: __________________
Settings held fixed: __________________________________________________
| Driver | Current evidence | Low assumption | Base assumption | High assumption | Confidence |
|---|
| Demand | | | | | Low / Med / High |
|---|
| Price or conversion | | | | | Low / Med / High |
|---|
| Capacity or cost | | | | | Low / Med / High |
|---|
| Forecast review | Value | Explanation |
|---|
| Low / base / high | ___ / ___ / ___ | Why is this range plausible? |
|---|
| Actual result | | Inside or outside the range? |
|---|
| Signed error | Actual − base = ___ | Positive or negative direction? |
|---|
| Absolute error | |Actual − base| = ___ | How large is the miss in context? |
|---|
| Guardrail actual | | Was a constraint threatened? |
|---|
| Next low / base / high | ___ / ___ / ___ | What evidence changed the forecast? |
|---|
Diagnose error without hiding it
Signed error: actual − forecast. A positive value means the forecast was low; a negative value means it was high.
Absolute error: |actual − forecast|. It compares miss size without canceling high and low misses.
Range coverage: record whether the actual fell between low and high. Repeated misses on one side may indicate bias.
Percentage error: |actual − forecast| ÷ |actual| × 100% can help compare scale, but is undefined when actual is zero and unstable near zero. Always report the original unit too.
Check forecast quality
- Was the time period and unit explicit?
- Did each scenario have a stated driver assumption?
- Was the forecast committed before the actual appeared?
- Did opposite errors cancel in an average?
- Are misses repeatedly high or repeatedly low?
- Did capacity, quality, cash, or safety limit the result?
- Does the update use new evidence rather than a preferred story?
A wider range is not automatically more useful. It should be narrow enough to guide a decision and wide enough to reflect supported uncertainty.
Write a forecast update that supports a decision
- Prior forecast: state the low, base, high, period, and unit.
- Actual: report the result, signed error, absolute error, and range coverage.
- Diagnosis: name the assumption that best explains the miss and distinguish evidence from speculation.
- Next forecast: provide the revised range and show what changed.
- Action: connect the update to inventory, staffing, capacity, cash, or another bounded decision.
- Trigger: state the next observation that would cause another update.
Sentence frame: We forecast ___–___ with a base of ___ for ___. Actual ___ was ___ above/below base and inside/outside the range. Because evidence about ___ changed, the next base is ___. We will ___ while protecting ___, and review again when ___.
16-point forecasting lesson rubric
| Criterion | 4 — strong | 3 — capable | 2 — developing | 1 — beginning |
|---|
| Forecast design | Clear period, outcome, unit, decision, guardrail, and coherent range. | Complete with one minor ambiguity. | Several details or range logic missing. | Unsupported single guess. |
|---|
| Assumptions | Drivers are explicit, evidence-linked, classified, and confidence-rated. | Relevant assumptions with partial evidence. | Assumptions are vague or mixed with facts. | No usable assumption record. |
|---|
| Error diagnosis | Accurate error, coverage, direction, bias check, and guardrail review. | Accurate core calculations and plausible diagnosis. | Calculation or interpretation gaps. | Miss hidden or original changed. |
|---|
| Rolling update | Evidence-based range, bounded action, trigger, and model limitation. | Reasonable update with most elements. | Update weakly tied to evidence. | No defensible revision. |
|---|
Reward transparent learning from misses, not closeness alone. Use the full simulation assessment rubric for collaboration and feedback codes.
Teacher notes, support, and extension
Prevent hindsight edits
Have pairs initial or photograph only their paper forecast before running. Students should preserve the original, then add a dated update beside it.
Support learners
Provide one approved outcome, three suggested drivers, and starter low/base/high values. Ask students to explain the sign of error verbally before calculating percentages.
Extend the analysis
Run three rolling periods, graph forecast and actual, calculate mean absolute error, and audit whether errors cluster on one side. Compare a simple baseline forecast against a more complex method.
Forecast responsibly, privately, and safely
Simulation output comes from simplified educational rules. It is not market research, a demand signal, financial advice, or evidence that a real product, price, staffing level, or capacity plan is lawful or safe. Do not invent customer data or collect student names, contact details, precise locations, health information, or other personal information for this activity.
A real forecast requires current and reliable source data, appropriate permissions, legal and regulatory review, safety and accessibility constraints, stakeholder input, and documented methods. Never manipulate a range to justify a preferred decision, promise an unsupported result, or remove an inconvenient miss.
Frequently asked questions
How does a simulation teach forecasting?
Students forecast from explicit assumptions, test the forecast, measure error, and revise the next period.
What should students forecast?
One measurable decision outcome plus a capacity, quality, safety, or cash guardrail.
What is forecast error?
Actual minus forecast shows direction; absolute error shows miss size.
Why use low, base, and high?
A range makes uncertainty visible while preserving a base case for planning.
Can this predict a real business?
No. It tests reasoning inside a simplified educational model.