Supply Chain · Forecast
Monthly Demand Review
Compare forecast and actuals item by item, flag the big misses, and capture the reasons before re-forecasting.
- Time1 h
- FormatTeam
- StageForecast
Monthly Demand Review: what it is and why it works
The monthly demand review is a structured look back at last month's forecast against actual demand, followed by a decision on how to adjust the next forecast. The team ranks items or families by absolute forecast error, focuses on the ten largest misses, and writes one sentence per miss explaining its real cause. Causes are then sorted into three buckets: data problems such as late orders or wrong units, model problems such as a missed trend or seasonality, and one-off events such as a promotion, a strike or a large project order. Only recurring causes feed into next month's forecast adjustments.
The review works because forecast errors are rarely random at the top of the list. The biggest misses usually have explainable causes, and each explained cause is a chance to improve inputs, models or communication with sales. Focusing on ten items keeps the meeting short and the analysis deep, instead of skimming hundreds of lines. The review is the first step of an S&OP cycle, provides the qualitative context behind the numbers in a forecast accuracy audit, and generates the feature ideas, such as promotion flags or customer project calendars, that make statistical or machine learning forecasts more useful.
What you need
- Last month's final forecast and actual demand by item or family, at the planning level used for decisions
- A ranked list of absolute errors, prepared before the meeting
- Participants from demand planning, sales or account management, and supply planning
- Notes on known events: promotions, price changes, customer projects, supply cuts
What you get
- A one-sentence cause for each of the ten largest misses
- Causes classified as data, model or one-off event
- Specific adjustments or corrections for next month's forecast
- An action list for data fixes and model changes, with owners
When to use it
When forecasts are produced but never interrogated.
How to do it, step by step
- Pull last month’s forecast and actuals per item or family.
- Rank items by absolute error and focus on the top ten only.
- For each, write the actual cause of the miss in one sentence.
- Separate causes into data, model, and one-off events.
- Feed the recurring causes into next month’s forecasting adjustments.
Worked example: Monthly review at an industrial fastener distributor
Illustrative scenario — figures are realistic but not from a real company.
A distributor of industrial fasteners and fittings forecast about 2,500 active items for three regional warehouses. Monthly weighted error hovered around 38%, and the planning team spent most of its time adjusting numbers without knowing whether the adjustments helped.
- The planner ranked last month's error by absolute units times unit cost and brought the top ten families to a one-hour meeting with two sales managers.
- For each miss the group wrote one sentence. Stainless flange bolts were over-forecast by 40% because a mining customer had delayed a shutdown; a hydraulic fitting family was under-forecast because a new OEM account had not been entered.
- Classifying the ten causes gave four data issues, three model issues and three one-off events.
- Data issues led to a rule that new accounts and shutdown schedules go into the planning system within a week of being known. Model issues showed that two families had clear seasonality that the moving average ignored.
Result. Over the next quarter weighted error fell to about 29%, mostly from the new-account and shutdown-calendar rules. The team learned that the most valuable output of the meeting was not the adjusted numbers but the fixes to inputs that prevented the same miss from happening again.
Common pitfalls and how to avoid them
- Reviewing every item in the portfolio so the meeting becomes a line-by-line recital.Limit discussion to the top ten errors by value or units and prepare the ranking before the meeting.
- Accepting vague explanations such as "the market was slow."Require one specific, checkable sentence per miss, naming the customer, event or data problem.
- Correcting one-off events as if they were trends.Separate one-off causes from recurring ones and adjust the model only for patterns that repeat.
- Letting sales overrides flow into the forecast without tracking whether they improve it.Record overrides and compare their accuracy to the statistical forecast in later reviews.
Frequently asked questions
What is a demand review in S&OP?
It is the step in the monthly Sales and Operations Planning cycle where sales, marketing and demand planning agree on an unconstrained demand plan. It usually starts with a look back at forecast performance and the reasons for major misses, then updates assumptions and the forecast for the planning horizon. The result is passed to the supply review, where capacity and materials are checked against it.
Who should attend a monthly demand review?
The demand planner leads, with sales or account managers who know customers, marketing if promotions or launches matter, and a supply planner to hear the implications early. Keep the group small enough to discuss causes openly. Senior leaders usually join later in the S&OP cycle rather than in the demand review itself.
How do you analyze forecast error causes?
Rank errors by absolute size, ideally weighted by value, and pick the largest. For each one, ask what actually happened: an order that was late or early, a customer change, a data entry problem, or a pattern the model did not capture. Write the cause in one sentence, then group the causes to see which ones recur. Recurring causes point to process or model fixes.
Origin
S&OP component — Oliver Wight methodology, 1980s.
Used in these playbooks
S&OP setup quarter 1 quarter
Install the monthly S&OP rhythm in one quarter: demand review, supply review, one reconciliation with real trade-offs.
Related methods
- Forecast Accuracy AuditMeasure MAPE and bias per item family; a systematic bias means the process, not the model, is broken.
- S&OP Meeting DesignSet up the monthly Sales & Operations Planning rhythm: demand review, supply review, and one reconciliation…
- ML Forecasting PilotStart one machine-learning model on one product family, benchmark it against your current method for three…
More in “Forecast”
Anticipate demand so the chain plans from facts, not hope.