S&OP Preparedness: The Solution Is Not in the Forecast

In planning and S&OP discussions, one hope is often expressed in different ways: if only we can find the right forecast, everything will start working. The hope is understandable. The problem is that it is wrong and expensive. The real issue lies elsewhere: the organization has not agreed in advance on what to prepare for and how to respond to exceptions. The value of planning is only realized when it guides decisions related to customer commitments, capacity, inventory, and procurement.

A second problem becomes visible in planning meetings. Time is spent explaining and reconciling report figures when it should be used for decisions: what should be done if demand exceeds or falls below plan, and how should supply chain disruptions be managed? A report tells you what happened, not what should happen next.

Since the COVID era, a third observation has been repeatedly raised in executive and planning teams: forecasts no longer work the way they used to.

The observation is correct. The conclusion often drawn from it is not: we need better forecasts. A better forecast alone will not solve the problem, and pursuing perfect forecast accuracy can be an expensive distraction. How could things be approached differently?

During the Apollo 13 mission, an oxygen tank exploded more than 300,000 kilometers from Earth. The original flight plan turned into a rescue operation within minutes. In mission control in Houston, the decisive factors were a shared situational picture, problem-solving, collaboration, and the ability to make decisions under pressure. The same applies to S&OP and business decision-making: a forecast will not always come true, and uncertainty cannot be eliminated, but it can be anticipated and managed.

A Forecasting Engine Does Not Replace Thinking

When forecasts fail, the first reaction is often to acquire a better forecast: a new tool, algorithm, or system. The solution is appealing because the purchasing decision can be made quickly and in one step. Fixing planning process issues is slower and more difficult.

The problem is that a forecasting engine does not replace thinking. It produces a number, but it does not decide what actions should be taken based on that number, who should make the decision, or when action should be taken.

In practice, a forecast may exist, but it does not guide anything. Production priorities are decided separately. Inventory parameters are not updated accordingly, and purchase proposals are still generated using outdated lead times. In that case, the forecast is a report, not a control signal.

This is where technology investments start to look attractive. They provide a quick way to demonstrate action without touching the process itself. The motivation is not limited to management. The pressure to adopt AI also affects IT and business functions, for whom a technology investment offers a visible response to expectations. The initiative moves forward and all parties benefit in the short term, but no one asks what was actually broken in the process.

I recognize the temptation myself. I solved my golf putting problems this summer by buying a new putter, and the new club appears to have fixed everything. At least from the salesperson’s perspective. And if it does not solve the problem, then obviously the issue is the wrong tool, so next summer I will buy another putter.

The return on a forecasting investment does not come from a better forecasting model. It comes from the decisions connected to the forecast. If that connection to decision-making is missing, a more accurate forecast will neither improve delivery performance nor free up working capital.

Forecasts Have Not Stopped Working. Decision-Making Has Fallen Behind

Forecasting methods have not deteriorated. The operating environment has changed faster than companies' ability to respond. Historical data describes a market that no longer exists. Disruptions and delays have increased, and variability in lead times has often grown even more than variability in demand.

The most concrete example relates to availability. A supplier informs you: “We delivered to another customer because they placed a binding order first. We can no longer deliver to you.” Deliverable quantities are determined when an order is placed, not when a forecast is created. A forecast that has not been converted into a reservation or commitment does not protect capacity or availability. Available-to-promise risk, the risk of losing promised availability, becomes increasingly important as companies outsource larger portions of their supply chains.

This is the point many overlook. The same pattern repeats throughout the chain: suppliers prioritize customers who have secured capacity or placed binding orders, and companies behave the same way toward their own customers. If capacity is not reserved and rules are not agreed upon, delivery performance becomes dependent on someone else's ordering cycle and priorities.

The same structure exists in master data and planning parameters. An incorrect product structure is rarely an isolated error. It is often a symptom of a broader problem where planning parameters, lead times, and units of measure are also flawed. Fixing a single error does not improve control if the next error is generated by the same source. The same applies to forecasting: a forecast error may be the symptom, but poor source data and parameters are often the cause.

Few organizations measure the time from detecting a change to making a decision, and from making a decision to implementing it. If that period is measured in months while the operating environment changes in days or weeks, forecast accuracy becomes a secondary metric.

Outsourcing often reduces costs and tied-up capital, but it also transfers part of delivery control to the supplier. In that case, customer commitments depend on supplier priorities, capacity, and ordering cycles. Costs can be outsourced, but delivery promises cannot.

Forecast Accuracy Has a Ceiling That No Model Can Exceed

Forecast accuracy is a misleading metric without context. The same figure can mean very different things depending on whether you are looking at a single item or a product group, the forecasting horizon, and the sales volume involved. Stable, high-volume demand is generally easier to forecast than low-volume, long-term, or seasonal demand.

The key factor is demand variation relative to the mean, commonly referred to as the coefficient of variation. If demand fluctuates as much as or more than its average sales volume, a reliable forecasting model cannot be built from historical data until the demand structure, sources of variation, and exceptions have been analyzed separately.

The classification thresholds are practical conventions, but the underlying mathematical limitation is real. If demand fluctuates randomly around its average, the best available forecast may simply be the average itself. When demand varies as much as average sales, meaning the coefficient of variation is 1.0, the theoretical upper limit of forecast accuracy may only be around 20–30%, far below 80%. This leads to two important observations.

First, a goal such as “increase forecast accuracy to 85%” may be practically impossible for highly variable items. Before setting targets, organizations should review demand variability, aggregation level, forecast horizon, and performance metrics. Otherwise, money is being spent on something that cannot be achieved.

Second, an accuracy figure is meaningless without understanding the aggregation level. When twenty independent products are aggregated into a single group, demand appears much more predictable, even though no operational decision-making has improved. Reported forecast accuracy can therefore increase simply by changing the level of analysis.

A clarification is important. Stable, high-volume demand does not require analyst time or expensive computing resources if the demand pattern repeats year after year and orders arrive continuously. In those situations, steady demand is a safe assumption, and the supply chain can be managed through simple demand-driven replenishment rules.

In demand planning improvement initiatives, one question should always be asked: how have exceptions been combined with the stable baseline demand? Have seasonal fluctuations, campaigns, and known customer events been systematically modelled? If not, a better tool and improved model may genuinely help. There is little value in investing to forecast an obvious baseline slightly more accurately.

The same question should be asked on the supply side. How have production and subcontracting constraints and exceptions been considered? Have production bottlenecks, supplier capacity limits, maintenance shutdowns, and delivery disruptions been modelled as scenarios? If the ability to model capacity constraints and resolve bottlenecks is missing, an advanced planning solution may quickly pay for itself. For managing routine operations, however, a new system is unnecessary. It is like bringing a tank to a fistfight.

Bram Desmet’s service-cost-cash triangle turns forecasting limits into a management question. Service performance, cost, and working capital cannot all be optimized simultaneously. Trade-offs must be made consciously. When forecast accuracy reaches its practical limit, uncertainty must still be paid for, either through inventory, flexible capacity, or lost sales. One of the three is unavoidable. Management's responsibility is to decide in which form the cost will be accepted.

A local inventory buffer for high-margin or business-critical products can be a financially justified solution. In such cases, losing even a single sale may be more expensive than holding the product in stock. The key question is therefore which costs the business more: additional inventory or lost sales? When the financial impact of stock-outs exceeds the cost of carrying inventory, delivery performance should be protected with a larger local buffer. The same calculation should be applied to expedited deliveries and air freight. In these situations, supply chain optimization is not based on achieving the highest possible forecast accuracy, but on selecting the most economically sound overall solution and maximizing business performance.

Instead of pursuing forecast accuracy, organizations should first distinguish between the portion of demand that is genuinely predictable and the portion that is not. Only then can they decide what should be managed through pull-based replenishment, what should be protected with inventory buffers, and where flexible internal or subcontracted capacity and jointly agreed operating rules are required.

A smaller inventory is not always a better inventory. The financial impact of stockouts involving high-margin or business-critical products can be greater than the cost of capital tied up in inventory. For this reason, the primary objective of inventory management should not be minimizing working capital, but optimizing overall business performance.

S&OP Preparedness Is a Planning Decision, Not a Reaction

All planning is based on assumptions about the future. The difference between good and poor planning is not how many scenarios materialize, but how well potential deviations have been anticipated in advance.

The following four practical examples illustrate how preparedness is transformed into decisions made before problems arise.

Shifting Production Priorities
Production had previously focused on high-volume products. The priority was shifted to make-to-order products, while inventory for high-volume products began to be replenished based on demand. The logic was simple: high-volume products can tolerate inventory, whereas customer-specific order products cannot wait. Capacity was allocated where flexibility was most limited.

Building Flexibility into the Production Schedule
The company's production plan had previously been fixed and inflexible. A free production slot was intentionally reserved at the end of each week for urgent requirements. The rest of the schedule could be frozen because flexibility had been built into the plan itself. Spare capacity was continuously available regardless of whether the need had been forecast in advance.

Managing the Product Portfolio
The company had previously expanded its product range year after year without a clear method for phasing out older products. The number of products increased, as did the capital tied up in inventory. The company divided products into stocked items and make-to-order items, defined clear operating models and planning parameters for each category, and reclassified some previously stocked items as make-to-order products. Uncertain demand was no longer served through expensive local inventory. Inventory value decreased by more than 30%, while product availability improved. The company also had significantly more cash on hand than in previous years.

Shifting the Inventory Buffer
The company held raw materials, semi-finished goods, and finished products throughout the production chain. This tied up capital and made deviations more difficult to detect. Target levels, control values, and performance metrics were defined for different stages of production, and the inventory focus was shifted from raw materials and finished products to semi-finished goods. This allowed the company to respond more quickly to demand fluctuations. During the first year, finished goods inventory value decreased by more than 20%, transparency across the production chain improved, and manual spreadsheet work was reduced.

None of these solutions improve forecast accuracy. Yet every one of them either improves delivery capability or frees up capital. The conclusion is clear: uncertainty is not managed solely through a smarter algorithm, but through better operating principles and a more effective structure.

This is where the perspective changes. The challenge is no longer limited to forecasting demand for an individual product. It is about managing capacity across the entire value chain. The question is no longer just what we can sell, but what we are capable of responding to.

Executive leadership should ask the planning organization two questions: How quickly can we respond to a last-minute change, and what does that level of readiness cost?

Scenarios Must Be Decided Before Things Go Wrong

Preparedness is not an abstract objective. It is a set of decisions agreed upon in advance. Supply chains repeatedly face the same four situations: demand exceeds or falls below plan, procurement exceeds demand, or delivery delays and disruptions occur. The critical question is whether the response model has been agreed upon before the deviation actually happens.

Demand Exceeds Expectations. Companies should decide in advance what inventory buffers are maintained, which customers they are reserved for, and how scarce supply will be allocated between customers. Without predefined allocation rules, decisions are made under pressure, and the most important customer may not receive products first.

Demand Falls Below Expectations. Organizations should determine beforehand how excess inventory will be managed: who the products can be sold to, through which channels, and how much pricing flexibility is acceptable. These decisions should not be postponed until products are already becoming obsolete and write-down risks are imminent.

Procurement Exceeds Demand. Control mechanisms should be agreed upon in supplier contracts and collaboration models ahead of time. This makes it possible to react when demand weakens while long lead-time purchase orders continue to arrive. The risk is particularly significant in country-based organizations that must purchase from global suppliers in volumes that exceed local demand. Supplier agreements should address flexible delivery quantities, postponement conditions, order modifications, and phased deliveries. This helps prevent receiving docks from becoming overloaded and inventory from unnecessarily tying up working capital.

Delivery Delays and Disruptions. Organizations should agree in advance on how to respond if a primary supplier is delayed or delivers only partial quantities. Alternative sourcing channels and allocation rules for critical materials within production should be established before disruptions occur. Otherwise, components end up being sourced from spot markets only after the production line has already stopped.

If an organization reacts slowly, individual exceptions begin to repeat and eventually become a permanent business problem. The key word here is chronic. What starts as an isolated issue becomes the normal state of operations, and valuable time is spent patching problems and explaining failures instead of creating business results. The financial consequences are direct. Excess inventory ties up working capital and increases the risk of write-downs, while supply shortages reduce revenue and weaken customer satisfaction. Ultimately, the impact becomes visible where it matters most: EBITDA and cash flow.

In its 2024 evaluation, Gartner identified scenario management and process management as key differentiating capabilities of modern planning solutions. At the same time, most companies were assessed as having only low to moderate planning maturity. The need has been recognized, but practical capability continues to lag behind.

One Number Is No Longer Enough

Sales, production, procurement, and finance often pull in different directions, and that is no coincidence. The conflict is usually rooted in deeply embedded ways of working and in how different functions are measured.

- Sales is focused on growth and product availability.
- Production seeks efficiency through long production runs.
- Procurement aims for lower prices through larger order quantities.
- Finance seeks to reduce tied-up capital.

Each objective is rational on its own, but without a shared framework for decision-making, they often collide and undermine both operational efficiency and business performance.

This explains why the forecast so often becomes a convenient scapegoat. Each function views the problem through the lens of its own objectives and looks for the root cause outside its area of responsibility. The forecast becomes an easy target because no single function owns it. As a result, the forecast is blamed for problems that are actually caused by conflicting objectives or the absence of agreed operating principles, not necessarily by forecasting itself.

"One number" does not simply mean a single number. It means the ability to make decisions based on a common understanding of the business. Achieving this requires bringing often conflicting functional objectives to the table and resolving the associated trade-offs together. A common number alone is not enough, however, because the operating environment often changes faster than the plan itself. Alongside it, organizations need agreed response models: what do we do if demand, availability, or capacity falls short? These decisions belong on management's agenda. A planner alone cannot reconcile conflicting metrics or resolve the trade-offs that pull the organization in different directions.

For this reason, S&OP development should begin with process readiness and scenario readiness, not with technology. Technology readiness is routinely assessed in development initiatives, and data readiness is increasingly being evaluated as well, but process readiness often remains overlooked. Before selecting or implementing a tool, organizations should understand what decisions are made within the process, who owns those decisions, and how exceptions are handled. An unclear process does not become clear simply through digitalization if insufficient attention is given to its design.

Before investing in a planning solution, organizations should first establish common operating principles and ensure that every key decision has a clearly defined owner, objective, agreed decision window, and response model. Only then does it make sense to assess what technology should support or automate. Without a solid foundation, even the best system will produce nothing more than better-formatted reports with little impact on decision-making.

Before the Next Technology Investment

A company does not lose to its competitors because a forecast was off by 5%. More often, it loses because it responds to change more slowly than its competitors. That is why management should evaluate the impact of any future technology investment carefully: will it improve decision-making, operational control, and responsiveness, or will it simply produce more accurate numbers?

If demand were to exceed the plan by 20% next month, would you already know:

- where the additional capacity or inventory would come from
- who would get less as a result
- how scarce supply would be allocated among customers
- who would make the decision

If the answer is no, the problem is not the forecast.

Conclusions

Forecast accuracy has a ceiling set by demand variability. Once that limit is reached, management must shift its focus from fine-tuning models to preparedness and responsiveness. Switching forecasting models will no longer solve the problem.

S&OP preparedness is a planning decision. Inventory buffers, contingency plans, and flexible capacity must be decided in advance based on what the company wants to protect and where it is willing to be flexible. Otherwise, decisions will be made too late and only under pressure.

Responsiveness is a measurable capability. Management should monitor two key delays: the time from detecting a change to making a decision, and the time from making a decision to execution. These metrics reveal far more about organizational performance than a single forecast accuracy percentage.

S&OP is a management rhythm, not a reporting meeting or a planning tool. Its effectiveness should be evaluated by whether meetings produce clear decisions, whether those decisions have owners and agreed implementation timelines, and whether they are actually executed.

Technology is a tool, not the objective. Its value is only realized when processes, decisions, and accountability are already in place. Management's first responsibility is to define what technology should support and what should be automated. In today's AI hype, this is often overlooked.

Uncertainty is not managed with a single better number. It is managed through decisions made in advance, organizational responsiveness, and the execution of those decisions.

Frequently Asked Questions

We have compiled the most frequently asked questions about S&OP, forecasting, preparedness, and planning development.

What does S&OP mean in practice?

S&OP is a controlled and repeatable planning process with designated owners, agreed roles, and defined metrics. It aligns the perspectives of sales, production, procurement, and finance. A practical measure of its quality is simple: how many decisions are made during the S&OP meeting.

Does successful S&OP require a perfect forecast?

No. Forecast accuracy has a ceiling determined by demand variability, and a better model cannot exceed it. Successful companies are distinguished not by forecast accuracy, but by their ability to evaluate alternatives, make decisions quickly, and respond to change.

What is a good level of forecast accuracy?

There is no universally applicable answer because the achievable level depends on demand variability, the level of aggregation, and the time horizon. As a rule of thumb, the higher the coefficient of variation, the lower the ceiling. For highly variable items, a target of 80% is mathematically impossible.

What does forecast accuracy actually mean?

A forecast accuracy figure is meaningless without context. The level of aggregation, time horizon, and volume all have a decisive impact on it. The same data produces very different accuracy figures at SKU level and product family level. Figures from different industries cannot be compared directly.

What is the difference between S&OP and IBP?

S&OP typically covers a 6 to 24-month planning horizon at volume level and balances demand and supply. IBP extends the same approach to a 24 to 36-month horizon and integrates volume with financial performance, including revenue, margin, and EBITDA. In IBP, finance often leads the process.

Where should S&OP development begin?

With process readiness, not with a tool. In practice, this means appointing owners, agreeing on a decision-making rhythm, and defining two scenarios in advance: what to do when demand exceeds the plan and what to do when it falls below the plan. Automating an unclear process only multiplies the uncertainty.

Is Your Challenge Forecasting or Preparedness?

Many companies continue investing in forecasting improvements when the greatest opportunity lies in decision-making, processes, capacity management, or supply chain responsiveness.

Let's review the most critical risks and bottlenecks in your supply chain, assess their business impact, and identify the best ways to prepare for them before they become costly problems.

Book a Consultation →

Lue seuraavaksi

ERP and AI Projects Rarely Fail Because of Technology. They Fail Because of Data. → Data Control → Sales & Ops →

Tommi Hyyrynen, CEO, Lanttu.io

Jouko Laakso, Schedio Oy, Senior Consultant / CEO