Sales and Operations Planning (S&OP) is a corporate practice intended to deliver superior supply chain execution by leveraging a deeper alignment with other divisions beyond supply chain - most notably sales, finance and production. The practice usually revolves around a monthly process starting from the sales forecasts and ending with quantified production plans. This practice emerged in the 80’s, along with ERPs and MRPs that provided the numbers to base the forecasts on.
The origin and motivation behind S&OP
The postwar economy of the 50’s and 60’s was, in multiple ways, straightforward: keep offerings narrow, increase production throughput, lower prices through economies of scale, and finally increase demand through mass media. However, by the end of the 20th century, supply chains had outgrown this model: more extensive product ranges, more geographic locations, more echelons. As a result, numerous inefficiencies had appeared, and the notion of supply chain emerged as a distinct practice from logistics. In this context, S&OP was coined in the 80’s as companies started to realize that internal disalignments were enough to generate substantial financial overheads. Both S&OP 1 and informational silos 2 were formalized in 1988.
At a symptomatic level, in the 80’s, large companies realized that a series of problems had become prevalent in their supply chains:
- at the same time, the company could face both sizeable overstocks and low service levels.
- oversized engineering teams ended up scrapping most R&D initiatives but remained late at delivering products that matter.
- marketing teams ended up amplifying both overproduction or underproduction problems by channelling their efforts to the wrong products.
Confronted with these problems, S&OP introduced a twofold distinctive answer. First, the company-wide alignment from early R&D to marketing pushes had to become the top management’s direct responsibility - CEOs included. This alignment would be created by following a specific process outlined by S&OP. Second, the process would explicitly and quantitatively become data-driven, a relative novelty of the late 80’s that had become possible as electronic stock levels and electronic stock movements were becoming mainstream through the adoption of ERPs.
The 5 steps of S&OP
The S&OP process is cyclical, going through a series of steps, every year, quarter and/or month depending on the choices made by the company. The CEO of the company is expected to own the process and make sure that various stakeholders dedicate enough resources to the S&OP initiative in order to deliver the intended company-wide benefits. The process is supposed to go through the following steps:
- Sales forecasting: The historical sales data as well as the quantitative insights from the sales teams are consolidated, typically following a bottom-up process, starting with sales employees. Raw demand forecasts are produced.
- Demand planning: Assessment and validation of the demand forecasts. Addition of structural insights about future demand, and identification of strategic risks, which may not be reflected by the raw forecasts, such as sources of expected variability (e.g. marketing actions).
- Supply planning: Assessment and validation of the projected capacities required to fulfill the demand, taking into account the projected variabilities both on the demand side and on the supply side. Prioritization and scheduling of the required operations.
- Reconciliation of plans: Reconciliation of the demand plan against the supply plan, and assessment of the company’s overall financial performance (gross margins, cash flows, long term customer retention, etc).
- Finalization of plans: Finalization of the plan and publication to make it widely accessible within the company and let the parties involved proceed with their respective contributions to the plan.
The S&OP process involves a series of meetings intended to foster focus, alignment and synchronization among all the functions of the organization. Those meetings are usually the opportunity to “re-plan”, leveraging the previous plan as the starting point of the discussion and driving the efforts where corrections are the most pressing.
At the software level, S&OP relies on the transactional backbone of the company - i.e. the ERP (Enterprise Resource Planning), the MRP (Material requirements planning), the WMS (Warehouse Management System), the TMS (Transportation Management System) - to obtain the relevant historical data, but it usually delegates the analytical workload to dedicated software components, typically an APS (Advance Planning and Scheduling). The APS explicitly supports S&OP both from a numerical perspective - to compute the statistical forecasts - and from a workflow perspective - to let users correct and validate the figures.
Despite the claims from multiple vendors that the “best-in-class” businesses operate under S&OP, most implementations suffer from similar flaws, which are intrinsic to the very nature of S&OP, namely:
- Some parties involved have structural incentives to distort the S&OP process in ways that cannot be countered without introducing other problems. For example, “sandbagging” refers to the widespread practice of putting forward highly conservative targets in order to “exceed expectations” later; which is typically the prime driver for promotion / bonuses within the company.
- The sheer number of parties involved in S&OP usually leads to “design by committee” situations where the company is incapable of taking decisive actions that may be essential to its survival, as those decisions may strongly antagonize many participants.
- Even in the most favorable situations, the S&OP process is invariably time-consuming for management teams within the company. The fact that the S&OP overhead is a necessary evil is debatable, but it’s always a heavy process.
- The forecasts are always incorrect to some extent and always a source of contention among parties. Attempts to improve the forecasting accuracy nearly always results in increased software complexity - at the expense of software reliability. Statistical forecasting tends to be opaque for most parties involved - including, frequently, the software vendor itself.
It is also notable that most criticisms - valid or not - voiced against S&OP are dismissed with the “No true Scotsman” fallacy. Philosophy professor Bradley Dowden gives the following simplified rendition of the fallacy:
Person A: “No Scotsman puts sugar on his porridge.”
Person B: “But my uncle Angus is a Scotsman and he puts sugar on his porridge.”
Person A: “But no true Scotsman puts sugar on his porridge.”
Indeed, in most companies struggling with their S&OP process, the consensus is that it’s their imperfect flavor of S&OP which is at fault, rather than considering the alternative perspective: while S&OP might be a necessary ingredient to operate the company, it comes with predictable downsides.
The limits of S&OP
Like most ideas, S&OP is a product of its time: the 80’s. Since this decade, the practice of the predictive optimization of supply chains has evolved in ways that weren’t fully conceivable at the time. Therefore, it could be argued that:
- S&OP emphasizes a simplistic perspective of the “future”, namely classic time series forecasts trying to reflect the expected future demand. Probabilistic forecasts do not exist in S&OP. Tail risks, associated suppliers or competitors, are not part of the model either.
- S&OP is slow because it emphasizes a “humans-in-the-loop” perspective. Many companies never manage to operate the monthly flavor of S&OP and remain stuck with quarterly revisions of the plan. In contrast, modern supply chains now operate with machine-driven decisions delivered with negligible latencies (minutes or less).
- S&OP is not geared around vast interconnected applicative landscapes that include digital marketplaces both on the demand side and and on the supply side, where companies seek not only internal alignment but market-level alignment (e.g. leveraging competitive intelligence data).
- S&OP downplays the diseconomies of scale that weren’t fully understood in the 80’s and have grown substantially worse in a world where supply chains are now vastly more complex, not only on a physical level (more products, more echelons, more transporters, etc.), but also on an IT level (traceability, compliance, cyber risk, etc).
In conclusion, S&OP correctly identifies many challenges that remain at the core of present day supply chains, such as the need for company-wide alignment and the importance of data-driven decisions. However, the recipes offered by the processes usually referred to as S&OP are dated.
Lokad’s take on S&OP
Best practices are ever moving targets. Our overarching criticism is that S&OP isn’t accretive : human resources required by the S&OP process are consumed instead of being invested. Yet, supply chains are now driven by numerical recipes delivered via software systems. S&OP focuses on improving the end-results, which is a never-ending process as input data gets continuously refreshed. In contrast, present day approaches focus on improving the numerical recipes themselves - which typically involve various flavors of high dimensional statistics (e.g. machine learning) - and then, lets those numerical recipes operate without further manual interventions.