It is a situation I repeatedly encounter in projects: forecast accuracy is around 50%. The plan changes weekly – yet Supply Chain is expected to derive stable procurement, staffing and production plans from it.

The obvious demand is: “Sales needs to forecast more accurately.” The uncomfortable truth is that in volatile markets, forecast accuracy is not purely a Sales problem.

Forecast thinking vs. uncertainty management

Organizations often try to improve forecasting through additional forecast rounds, more Sales involvement, manual adjustments, new tools and more ambitious accuracy targets. This can help – but only when demand is fundamentally forecastable.

The underlying problem often remains: a forecast is treated like a reliable customer order even though it only describes one possible future.

The classic conflict

Business model: campaign-driven, high variety, short-term and volatile. Planning model: forecast-based, long lead times and rigid capacities. The result is predictable: forecasts change constantly, inventory rises, shortages remain, staffing changes at short notice, suppliers receive continuously changing requirements and Operations works in firefighting mode.

This is not necessarily a performance problem. It may be the wrong planning model.

The bullwhip effect is back

Promotions and short-term campaigns create demand peaks that may last only a few weeks. The customer orders more, Sales raises the forecast, Supply Chain plans additional volume, Procurement buys material and Manufacturing adds capacity. The promotion ends and sales fall back – sometimes below normal because customers or distributors stocked up beforehand.

The demand spike travels through the supply chain like a wave and can amplify the further a company is from the end customer.

How to recognize a forecasting problem

Typical signals are persistently low forecast accuracy, high volatility between planning cycles, systematic over- or underplanning, heavy manual intervention, no separation between baseline and promotions, no segmentation by item type, short-term changes without governance and rising inventory despite declining delivery performance.

The real question

The central question is not “How do we make Sales forecast more accurately?” It is: “How do we design a planning process that can manage uncertainty professionally?”

A practical approach

Separate baseline demand from events

Regular demand should be planned separately from campaigns, promotions, projects and large one-off orders. Only then can recurring demand be distinguished from temporary effects.

Segment the portfolio

Standard products, seasonal items, new products, spare parts and project items require different planning logic. ABC/XYZ and life-cycle segmentation provide a useful foundation.

Connect forecast and order intake

The medium-term forecast should be complemented by a short-term S&OE process. S&OP sets direction; S&OE evaluates current order intake, inventory, capacity and delivery capability.

Measure more than accuracy

Forecast Bias matters as much as accuracy. Persistent overplanning creates inventory; persistent underplanning creates shortages, premium freight and overtime.

Use ranges instead of false precision

A single forecast number often suggests certainty that does not exist. A base scenario, upper and lower demand ranges, explicit opportunities and risks and predefined response actions are often more useful.

Use digital tools selectively

Modern demand-planning systems can identify patterns, seasonality, outliers and structural breaks more effectively. But they do not replace clear process logic. A poor planning process does not automatically become better through digitalization – usually just faster.

Why forecast projects fail

They often start with the wrong expectation: more forecast accuracy instead of better decision capability. The objective cannot be to predict the future perfectly. It must be to recognize change earlier, assess impact faster and initiate the right countermeasures in time.

Conclusion

Forecast accuracy of 50% is not an isolated Sales problem. It is a signal that data, processes, responsibilities and planning logic may not be sufficiently aligned with the business model.

Excellence does not come from adding more forecast loops. It comes from a planning system that does not ignore uncertainty, but makes it manageable.