When the system may not be the problem
When forecasting, inventory or replenishment performance begins to cause concern, it is natural to question the planning system. The recommendations appear increasingly difficult to explain, planners regularly override the outputs, inventory accumulates in the wrong places and product availability does not appear to justify the amount of stock being held. Faced with these problems, replacing or upgrading the technology can seem like an obvious next step.
It may also be the wrong place to start.
Planning systems do not produce decisions independently. They apply forecasting methods, inventory policies, parameters and business rules to the data they receive. If those underlying assumptions no longer reflect the way the supply chain operates, the system can produce inappropriate recommendations while functioning exactly as it was designed and configured to function.
Before concluding that the technology is failing, a business should therefore distinguish between the system itself, the methodology contained within it and the parameters used to apply that methodology. This article considers why that distinction matters, how planning assumptions become embedded over time and how reviewing them against actual operational evidence can produce better decisions without necessarily replacing the system.
A planning system reflects the assumptions within it
Every planning system contains assumptions about how demand and supply should be managed. Some are explicit, such as service-level targets, supplier lead times, order quantities and review periods. Others are embedded within forecasting settings, safety stock calculations, replenishment rules, product segmentation and exception tolerances.
When these settings are first established, they may be entirely reasonable. A supplier lead time may reflect the information available during implementation. A service level may correspond with the commercial priorities of the business at that point. A particular forecasting method may suit the demand patterns represented in the product range, while order quantities and replenishment frequencies may reflect the warehouse capacity, supplier terms and transport arrangements then in place.
The difficulty is that the business does not remain static. Product ranges change, suppliers improve or deteriorate, customer expectations develop, demand patterns become more or less predictable and operational constraints move elsewhere in the supply chain. The methodologies and parameters within the planning system do not necessarily change with them.
Unless they are deliberately reviewed, the system may continue applying assumptions established in a different business environment. It is not malfunctioning when it does so. It is using the information and rules it has been given, which is why apparently poor system performance may be evidence of an assumption problem rather than a technology problem.
Correct calculations do not guarantee appropriate decisions
A calculation can be mathematically correct without producing an appropriate operational decision. This is an important distinction because the apparent precision of a system-generated recommendation can conceal weaknesses in the assumptions behind it.
Consider a supplier shown in the system with a four-week lead time. If actual replenishment regularly takes six weeks, the system may correctly calculate the required order using the value it holds, but the resulting recommendation will not adequately protect the business against the period it really needs to cover.
The same issue can arise with safety stock. The calculation may have been implemented correctly, but the resulting inventory will only be appropriate if its inputs and assumptions remain relevant. The service target, demand variability, lead-time variability, review frequency and treatment of unusual demand can all influence the outcome. A technically correct calculation based on outdated or poorly understood inputs can create too much inventory, too little protection or simply put stock in the wrong places.
Forecasting methods require similar consideration. A method that performs reasonably well for stable, frequently demanded products may be far less appropriate for intermittent demand, seasonal products, promotions or newly introduced lines. Applying one default approach across a varied product range can make the system consistent without making its recommendations consistently useful.
The question is therefore not simply whether the system has performed the calculation correctly. The more important question is whether the method, inputs and parameters remain appropriate for the decision the business needs to make.
Parameters can become permanent by default
Planning parameters have a tendency to become established facts within a business, even when they began as assumptions. A service level is agreed and gradually becomes treated as the service level. A supplier lead time is entered during implementation and remains unchanged. A forecasting method is selected as a default, while minimum order quantities and replenishment rules continue to be applied long after the commercial or operational circumstances behind them have changed.
Over time, the original reasoning can become difficult to trace. The business may know the value held within the system without knowing why that value was chosen, when it was last reviewed or what evidence continues to support it.
This does not necessarily mean that the parameter is wrong, but it does mean that the parameter should be open to challenge. A service-level target should reflect the importance of availability, the consequences of shortage and the cost of holding inventory. A lead time should represent a clearly understood planning requirement and be tested against actual supplier performance. Forecasting and replenishment policies should reflect the characteristics of the products to which they are applied rather than remaining in place simply because they are the system defaults.
Parameters should therefore be treated as planning decisions, not permanent elements of the software. When the circumstances behind those decisions change, the parameters should be reviewed with them.
Planner overrides provide evidence worth examining
Manual intervention is often presented as a problem in its own right. If planners regularly override the system, the immediate response may be to enforce greater compliance or conclude that users do not trust the technology. Neither explanation is sufficient without understanding what is actually happening.
Experienced planners may hold information the system does not possess. They may know that a supplier is experiencing difficulties, that a customer is likely to change an order or that a product is being replaced. In those circumstances, an override may be entirely appropriate.
Repeated patterns of intervention, however, can reveal more systematic issues. If planners routinely increase recommendations for a particular supplier, reduce quantities for a certain product group or disregard forecasts under recurring conditions, their actions may indicate that the system is missing information, applying an unsuitable methodology or relying on an inappropriate parameter.
The opposite may also be true. Analysis may show that some habitual overrides do not improve the eventual outcome and may introduce additional variation into the planning process.
The purpose of reviewing overrides is not to assume that the planner is right and the system is wrong, or that the system should always take precedence. It is to understand why they are reaching different conclusions. That difference can provide valuable evidence about where the planning requirement, system logic and operational reality are no longer aligned.
Historical evidence can help challenge existing assumptions
Planning methods and parameters should not be retained solely because they appear theoretically sound or have been used for a long time. Where suitable data is available, their behaviour can be tested against conditions the business has actually experienced.
Historical back-testing allows alternative forecasting methods, inventory policies or parameter settings to be applied to previous periods so that their potential effect can be examined. The analysis can consider the forecasts that would have been generated, the inventory that might have been required, the frequency and duration of shortages and the extent to which different settings would have improved or weakened the resulting decisions.
This does not recreate the past perfectly, nor does it guarantee future performance. Historical data may contain one-off events, data-quality problems and circumstances that would have been handled differently at the time. Nevertheless, back-testing provides a stronger basis for discussion than intuition alone because it exposes how a proposed approach behaves across real products, suppliers and patterns of demand.
It can also reveal that no single method performs best in every circumstance. One forecasting approach may suit stable demand, while another is more appropriate for seasonal or intermittent items. A uniform service target may produce unnecessary inventory for some products while providing inadequate protection for others. Segmentation can therefore be an important part of the review, allowing methods and parameters to reflect meaningful differences across the range rather than applying complexity indiscriminately.
Complexity should earn its place
A review of planning performance can create pressure to adopt more sophisticated calculations. In some circumstances, greater statistical or analytical sophistication will produce a meaningful improvement and should be seriously considered. However, complexity should not be treated as evidence of quality in itself.
A more complicated method may improve a statistical measure without materially improving the operational decision. It may also be more difficult to explain, maintain and challenge, particularly when the people using the output cannot understand why the recommendation has changed.
The appropriate comparison is not between a simple method and a sophisticated method in the abstract. It is between the decisions each method produces, the outcomes those decisions support and the effort required to maintain confidence in them.
That assessment should consider forecast performance, inventory investment, product availability, operational practicality and the ability of users to understand when an output is sensible and when it requires intervention. If additional sophistication produces a worthwhile improvement across those measures, the complexity has earned its place. If the benefit is marginal, a simpler and more transparent approach may be more valuable.
The objective is not to find the cleverest calculation available. It is to establish the most appropriate method for the decision being made.
A new system may inherit the same problem
There are circumstances in which replacing a planning system is entirely justified. The existing platform may lack the necessary functionality, handle integration poorly, constrain important processes or no longer meet the scale and complexity of the operation.
Even then, understanding the existing methodology and parameters remains important.
If the business moves its established service levels, lead times, forecasting assumptions and replenishment policies into a new platform without first challenging them, the new technology may reproduce many of the same outcomes. The interface may improve and the system may offer greater capability, but the quality of the decisions will still depend on how that capability is configured.
A parameter review is therefore not an alternative to system replacement in every case. It is part of establishing what any existing or future system needs to do. By separating technology limitations from methodology and configuration issues, the business can define the requirement more clearly and avoid carrying poorly understood assumptions into the next solution.
Challenge the assumptions before replacing the technology
When a planning system appears to be underperforming, the problem may lie in the technology, but that conclusion should follow analysis rather than precede it.
The system’s outputs reflect the data it receives, the methodologies it applies and the parameters through which those methodologies are translated into recommendations. If service levels, lead times, forecasting settings, safety stock assumptions or replenishment policies no longer reflect operational reality, changing the system alone will not necessarily improve the resulting decisions.
The appropriate starting point is to examine those elements separately. Review the methods being used, establish the basis for the principal parameters, compare key assumptions with observed performance, analyse repeated planner overrides and test credible alternatives against historical data. This creates the evidence needed to determine whether the business requires different technology, different planning logic or simply more appropriate configuration.
A planning system should not be judged only by whether it performs its calculations correctly. It should be judged by whether those calculations support appropriate supply chain decisions.
Sometimes the system itself is not the problem. It is doing exactly what the business has configured it to do.