Who should be in charge of sales forecasting? Governance Across Sales, Purchasing, Pricing, and the Supply Chain
Edouard Calliati
CMO - CRO
August 20, 2026
At most retailers, demand forecasting isn't done just once—it's done four times. Sales forecasts what it wants to sell, purchasing forecasts what it dares to order, and the supply chain forecasts what it can deliver at the lowest cost.
And pricing, almost always immediately thereafter, translates that same figure into pricing or promotional decisions. Four interpretations, four approaches, but only one real market to contend with.
Pricing isn’t a service that checks the forecast after an order has been placed—it’s the function that transforms it most quickly into a signal visible to the customer, even before a truck is loaded. Overlooking it in governance is like managing three out of four decisions and letting the one that’s most visible to the end customer slip by without any oversight.
This article does not address the mechanics of calculating a forecast—a separate guide in this series already covers that. It addresses the issue that, in practice, causes the most silent damage within a retail organization: who should be responsible for THE forecast, and how four departments—retail operations, procurement, supply chain, and pricing—each of which is correct within its own scope, end up collectively producing a result that no one chose.
Three services, three demand forecasts
A company that thinks it has only one sales forecast is almost always mistaken. In reality, three departments calculate different versions of the forecast, without necessarily realizing it.
Each service starts with a different constraint:
- Sales and marketing start with a goal to achieve—a revenue target, market share, or a commitment made to senior management. Structural bias: optimism, since no one builds a sales plan based on the assumption of stagnation.
- Purchasing starts with the opposite constraint—avoiding ending up with excess inventory that ties up cash and ends up on clearance. Structural bias: caution, since when in doubt, we order a little less, never a little more.
- The supply chain operates at a third level: that of physical flow—truck loading, warehouse capacity, and transportation costs. Structural bias: smoothing is done at the network level to optimize logistics, not necessarily at the store level, which reflects the fine details of commercial reality.
Three legitimate approaches. Three different figures.
And in the vast majority of organizations, there is no mechanism in place to address these issues before they take effect—a breakdown here, excess inventory there, or an end-of-month meeting that turns into a settling of scores.
In addition to these three perspectives, there is a fourth factor that is all too often overlooked in discussions of governance: pricing. A demand forecast almost always translates first into a pricing or promotional decision, even before it results in an order.
Ignoring pricing in forecast governance is like managing the cause without managing its most visible consequence for the customer. The price displayed on the shelf or online is often the first place where a forecast variance becomes apparent—long before inventory becomes available or a stockout is displayed.
Only some supply chain organizations have developed at least three of the five capabilities deemed essential for steering the future—including a truly cross-functional, enterprise-wide strategy, which is the least common of the five (Gartner, survey of 579 supply chain professionals, press release, February 18, 2025).
The Symptoms of an Ungoverned Forecast
An organization that produces four competing forecasts without realizing it does not notice this immediately. The problem manifests itself through symptoms that are often misdiagnosed as inventory issues rather than governance issues.
Symptom 1: Simultaneous stockouts and excess inventory across different categories
A retailer may, in the same week, experience stockouts in a category experiencing strong sales momentum—the store was right about the trend, but purchases didn’t keep up—and excess inventory in another category that the supply chain restocked based on a logistics plan that was out of touch with on-the-ground reality.
This isn't a problem of a forecast that's generally incorrect. It's a problem of three forecasts that were never compared with one another.
Symptom 2: End-of-month meetings that turn into disputes over numbers
Each department defends its metric by citing a different reason:
- Retailers are defending their lost revenue by pointing to supply shortages.
- Purchasing managers defend their cautious approach by pointing to the markdown rate from previous months.
- The supply chain defends its logistics decisions by pointing to the transportation costs it has avoided.
Everyone is right about their own metric, but no one is looking at the same baseline figure. As a result, the steering committee spends more time reconciling different interpretations of reality than it does deciding anything for the coming month.
Symptom 3: Forecasting becomes a political issue rather than a professional one
When there is no common framework, each department learns to defend ITS forecast rather than challenge the most accurate figure. The debate shifts from actual demand to who is right—a self-perpetuating dynamic that makes each subsequent cycle more difficult than the last.
Symptom 4: Pricing inherits all the biases, without being at the table
Pricing often silently absorbs these three biases without ever being linked to their trade-offs. A promotion launched based on an optimistic sales forecast—without challenging the volume assumption with the purchasing department—results in eroded margins.
This eroded margin does not appear in the financial statements until several weeks later—well after the pricing decision has been made. It is a problem just as costly as the first three, but one that is rarely identified as such in steering committees.
Why Every Department Is Right—and That's Exactly the Problem
The natural reaction to these symptoms is to look for someone to blame—the department that is “at fault.” This is a misdiagnosis.
Each function properly optimizes the metric against which it is evaluated. The problem is not the bias of each department; it is the lack of a mechanism for them to communicate with one another before a decision is made.
Sales / Retail
Assessed based on revenue and market share. Structural bias toward optimism —underestimating demand is tantamount to setting a limit on one’s own growth.
Shopping
Evaluated based on inventory turnover and markdowns. A structural bias toward caution —when in doubt, order the safest quantity, not the most likely one.
Supply Chain
Evaluated based on transportation costs and service levels. Structural bias toward smoothing —the grid that optimizes a truck is not the grid that reflects a store.
Finance / Management
Evaluated based on budget compliance. Structural bias toward rigidity —a figure approved in January remains the benchmark, even when the reality on the ground has already changed.
Pricing / Category Management
Evaluated based on margin and price-image. Structural bias toward direct translation —converts a forecast into a pricing or promotional decision without always challenging the underlying volume assumption.
None of these biases is an individual fault. They are the logical consequences of incentive systems that were never designed to work together.
Forecasting governance does not seek to eliminate these biases—it seeks to make them visible, to confront them, and to turn them into an explicit point of decision-making rather than an implicit source of conflict.
Pricing, in particular, pays a heavy price for the lack of this dialogue. It is the function that translates a forecasting discrepancy into euro amounts visible to the customer the fastest—even before a warehouse has made a move or inventory has become available.
additional revenue and margin when pricing and promotional decisions are managed in a coordinated and analytical manner, rather than in isolation from one another—the performance gap that most retailers leave on the table due to a lack of common governance between forecasting and pricing (McKinsey & Company, “Pricing in Retail: Setting Strategy,” April 1, 2015).
Who is responsible for what among sales, purchasing, supply chain, and pricing?
The question “Who should lead the sales forecasting process?” has a wrong default answer—“the department with the most data” or “the one who speaks up the loudest in meetings”—and a structurally sound answer: no one should lead it alone, but someone must be responsible for overseeing it.
| Job title | What she focuses on first and foremost | Typical bias in the forecast | Role in a Streamlined S&OP Process |
|---|---|---|---|
| Sales / Retail | Revenue, market share | Overestimating to avoid setting a limit | Drives the business vision, challenged by the facts |
| Shopping | Inventory Turnover, Markdowns | Underestimating to safeguard cash flow | Consulted on supplier feasibility, not as sole decision-makers |
| Supply Chain | Logistics costs, service levels | Adjust demand based on the truck's capacity, not the store's | Converts the forecast into physical flow and reports constraints |
| Finance / Management | Budget compliance, margin | Locks in a figure approved at the beginning of the year | Final arbiter in the event of a persistent disagreement |
| Pricing / Category Management | Turnover, margin, price-image | Converts volume into price without always challenging the hypothesis | Provide the translation of the forecast price, as agreed upon with the trade and finance departments |
To put it this way: a single baseline forecast circulates among these five functions—each enriches it with its own perspective, and none of them recalculates it on its own in a corner. This is the difference between true governance and mere coordination based on good will, which falls apart as soon as the timeline tightens. The most common misconception at this stage—treating this operational forecast as if it had to match the budget approved in January—deserves its own discussion: see the dedicated article in this report.
What a Single, Managed Forecast Actually Changes
The principle behind this approach is not new: it’s S&OP (Sales & Operations Planning), or its streamlined version—IBP (Integrated Business Planning) —in its most mature form. The central idea can be summed up in one sentence: a single baseline forecast, enriched and challenged by each function based on its expertise, rather than four independent forecasts that are attempted to be reconciled after the fact.
It's not primarily a question of tools—it's a question of processes and roles. But the effect, which has been documented, is real:
additional EBIT on average for companies with amature Integrated Business Planning process, compared to those without one—with service levels 5 to 20 points higher and missed sales 40 to 50 percent lower (McKinsey & Company, “The Transformative Power of Integrated Business Planning,” May 25, 2022).
The improvement does not stem from a more sophisticated statistical model—the departments involved were already using reasonable calculation methods, each on their own. The improvement comes from eliminating the delay and the disconnect between the time a demand discrepancy arises and the time the entire organization takes it into account.
Pricing is often the one element missing from this elimination of lead time. It is the function that most quickly reflects a forecasting discrepancy in the customer's experience—through a price, a promotion, or a discount.
And yet, it is rarely invited to the table where that figure is determined. As a result, the department most exposed to the consequences of an inaccurate forecast is often the one with the least say in challenging it early on.
GENIUS Predict: A Unique, Explainable Forecast Shared Across Departments
The GENIUS Predict module centralizes demand forecasting over several rolling weeks, using three scenarios— Conservative, Balanced, and Aggressive —rather than a single figure imposed on all functions. The “AI Explanation” section details the factors that influenced each forecast, allowing procurement and the supply chain to challenge the business scenario with data-driven arguments, not just intuition.
Implementing a Streamlined S&OP Process in Five Steps
A comprehensive S&OP process, with its formalized monthly committees and strict schedule, is often out of reach for a medium-sized retailer—and isn’t necessary to achieve the main benefit. A streamlined version, built around five principles, is sufficient to reconcile the four competing forecasts, including pricing.
Appoint a cross-functional arbitrator
One person—not a department—responsible for resolving disagreements between sales, procurement, supply chain, and pricing. Neither a judge and jury nor merely a meeting facilitator: a decision-maker with the authority to set a final figure.
Building a Single Data Repository
Sales history, out-of-stock items, promotions, prices: the same database powers all four reports. Functions based on different data cannot converge, no matter how well the meeting is conducted.
Setting a two-beat rhythm
A brief preliminary meeting where each department presents its analysis and assumptions, followed by an arbitration committee that resolves significant discrepancies—not a single large meeting where everything is discussed but nothing is decided.
Document the assumptions, not just the numbers
When the sales, procurement, supply chain, and pricing teams submit an estimate, they also explain why. A discrepancy becomes a factual point of discussion, not a power struggle between departments.
Link — without confusing — to the budget
A quarterly review compares the rolling operational forecast with the finance department’s annual budget commitment, without ever forcing one to match the other.
improved forecast accuracy and a 10 to 30 percent reduction in inventory for companies with a high-performance S&OP process—compared to those with a basic or nonexistent S&OP process (Bain & Company, “Good Sales and Operations Planning Is No Longer Good Enough,” April 12, 2016).
Booper has published a dedicated guide on the calculation methodology that underpins this shared framework—including the necessary data, the method for generating the figure, and accuracy metrics such as MAPE—titled “Sales Forecasting with AI: Methodology and KPIs.” This guide assumes that this figure already exists somewhere within the organization, often in multiple versions—and focuses on the question of which one should be considered authoritative, and who decides.
Four Common Misconceptions About S&OP
- “This is a process reserved for large corporations.” False: The streamlined version described above is based on five principles—not on specialized software or a full-time planning department. A medium-sized retailer can implement it in just a few monthly cycles.
- “We don’t need another meeting.” The problem isn’t the number of meetings; it’s their purpose. An arbitration meeting that resolves a documented discrepancy effectively replaces, in practice, several crisis meetings that merely assess the damage after the fact.
- “Consensus obscures useful signals.” Well-designed governance does not seek consensus—it seeks arbitration. Disagreement between sales and procurement on a particular category remains a valuable signal; it must be visible and debated, not quietly averaged out in a spreadsheet.
- “A single baseline forecast eliminates the nuances of each business line.” On the contrary: it is the lack of a common framework that forces each function to rely on its own estimate, since it cannot challenge those of others with shared arguments.
One shared forecast, not four competing forecasts
The sales, purchasing, supply chain, and pricing departments aren’t wrong when they each produce their own interpretation of demand—they’re doing their jobs, based on the metrics by which they’re evaluated. The problem is never the existence of these four assessments; it’s the lack of a space where they can be compared before they impact the store—a stockout, excess inventory, a poorly calibrated promotion, or a meeting that devolves into a dispute over numbers.
A streamlined S&OP process does not replace any of the existing areas of expertise. It provides them with a common framework, a regular forum for discussion, and an arbiter—and that is what transforms competing forecasts into a single, well-supported decision.
The broader topic of what happens to a forecast once it informs a decision is addressed in the flagship article of this special report, and the most costly consequence of poor governance—stockouts—is discussed in the dedicated article in this special report.
A Checklist to Determine Whether Your Organization Has a Problem with Forecasting Silos
- Do the sales, procurement, supply chain, and pricing teams work from the same source database, or does each team use its own export?
- Have you ever experienced, in the same week, a stockout in one category and excess inventory in another, with no apparent connection between the two?
- Do your end-of-month meetings focus on demand variances, or do they spend most of their time reconciling discrepancies in the numbers?
- Is there a specific person responsible for resolving a disagreement between two departments, or does the disagreement remain unresolved until the following month?
- Are the assumptions behind each forecast documented, or is only the final figure provided?
- Is pricing a factor in arbitrage, or is the forecast provided after the decision has already been made elsewhere?
Want to look beyond your organization's four competing forecasts?
Spend 30 minutes with our team to identify where, within your current process, sales, procurement, supply chain, and pricing are unintentionally out of sync.
Frequently Asked Questions
Who should be responsible for sales forecasting among sales, purchasing, and the supply chain?
None of the three departments should manage it alone: each brings legitimate expertise, but with its own structural bias— sales tends toward optimism (under-planning would be like setting a ceiling for itself), procurement toward caution (when in doubt, order the safest quantity), and the supply chain toward logistical smoothing, organized at the truck level rather than the store level.
This article introduces a fourth factor that is all too often missing from this governance process: pricing, which almost always translates forecasts into pricing or promotional decisions even before an order is placed—and thus inherits all the biases of the other three factors without being involved in their trade-offs.
The right structural solution is to appoint a cross-functional arbitrator—an individual (not a department)—responsible for resolving disagreements based on a common data repository shared by the four functions, in line with a pricing organization that has clearly defined roles.
It is this explicit comparison of hypotheses—not the elimination of biases themselves—that transforms four competing forecasts into a single decision embraced by the organization.
What is S&OP (Sales & Operations Planning)?
S&OP is a process that periodically brings together the sales, procurement, supply chain, and finance functions—including pricing, according to this article—around a single demand forecast, so they can compare their assumptions, which are based on a sufficiently deep sales history, and resolve discrepancies before they lead to stockouts, excess inventory, or inconsistent pricing decisions.
Its most advanced version isIBP (Integrated Business Planning): a single baseline forecast, enriched and challenged by each function based on its expertise, rather than four independent forecasts reconciled after the fact.
This article outlines a streamlined version based on five principles: an appointed cross-functional arbitrator, a single data repository, a two-step process (pre-meeting followed by an arbitration committee), documented assumptions, and a quarterly review that links the operational forecast to the budget without confusing the two.
A formal monthly committee meeting involving the entire group is not a prerequisite: this streamlined version is sufficient for most medium-sized retailers to reap the main benefit of the program.
Why might a retailer experience both stockouts and excess inventory at the same time across different product categories?
This is a typical symptom of an unmanaged forecast: Sales may have correctly identified a trend that Purchasing did not follow, leading to a stockout, while the supply chain restocks another category based on a logistics smoothing process that is disconnected from on-the-ground reality—often because the forecast failed to incorporate the right exogenous events—resulting in excess inventory.
This article emphasizes a key point: the problem isn't that the overall forecast was incorrect, but rather that the three forecasts were never compared with one another before they affected the store.
This problem often persists during end-of-month meetings: each department defends its own metric—sales focuses on stockouts, purchasing on its markdown rate, and the supply chain on transportation costs saved—without anyone looking at the same baseline figure.
Without a common framework, the steering committee ends up spending more time reconciling different interpretations of reality than deciding anything for the coming month.
Does a medium-sized retailer need a full-scale S&OP or a streamlined version?
In most cases, a streamlined version is sufficient: a designated cross-functional arbitrator, a single data repository, a two-step process (submission of assumptions followed by an arbitration committee), and documented assumptions rather than a single final figure.
This article debunks the common misconception that S&OP is reserved for large corporations: the streamlined version is based on five principles—not on specialized software or a full-time planning department—and can be implemented in just a few monthly cycles.
The available data confirms the value of the system even on this scale: a high-performance S&OP process yields a 20 to 50 percent improvement in forecast accuracy and a 10 to 30 percent reduction in inventory—an issue similar to the one addressed in our guide to forecasting ending inventory and seasonality, compared to a basic or non-existent process, according to Bain & Company.
A formal monthly committee meeting involving a large group is therefore not a prerequisite: it is the regular testing of hypotheses—not the cumbersome nature of the process—that yields the greatest benefit.
How can I tell if my organization has a problem with forecasting silos?
Some reliable indicators identified in this article: sales, procurement, and the supply chain operate on different databases, due to a lack of a clear distinction between internal and external data for pricing; end-of-month meetings focus on reconciling figures rather than on demand variances; and there is no explicit arbitration mechanism for resolving disagreements between two functions.
The checklist provided at the end of this article includes a factor that is often overlooked: Is pricing factored into the decision-making process, or is the forecast provided only after the decision has already been made elsewhere—a symptom just as revealing as simultaneous stockouts or overstocks?
Another concrete indication: having already observed, in the same week, a stockout in one category and an overstock in another, with no apparent connection between the two—a sign that forecasts are never cross-checked before they affect the store.
These issues cannot be resolved simply by switching forecasting tools; they require a common data repository and a cross-functional arbitrator, as detailed in the streamlined S&OP method described in this article.
What is the difference between S&OP and simple monthly reporting?
Monthly reporting identifies a variance after it has occurred: it measures, but it does not decide. An S&OP process—even a streamlined one—compares the assumptions of each function before an order or price is finalized, with a designated arbiter to make the final decision.
This article attributes this difference to the role of pricing: it is the function that most quickly translates a forecasting discrepancy into euros visible to the customer, through a price or a promotion—post-hoc reporting always comes too late to prevent the resulting erosion of profit margins.
The difference is also evident in the pace: the streamlined S&OP process involves a pre-meeting where each function presents its analysis and assumptions, followed by an arbitration committee that resolves significant discrepancies, rather than a single meeting where everything is discussed but nothing is decided.
It is this shift from measurement to decision-making that explains the documented gains: according to McKinsey, a mature IBP generates 1 to 2 additional points of EBIT and results in 40 to 50 percent fewer lost sales compared to simple post-hoc monitoring—gains that are managed through a pricing KPI dashboard shared across functions.
Also in this series
- Sales Forecasting: Methods, AI, and Best Practices (The Essential Guide)
- Sales Forecast vs. Budget: Why These Are Two Different Processes
- Out-of-Stock Situations: The Direct Link to Poor Sales Forecasting
Sources
- Gartner, " Gartner Survey Shows Only 29% of Supply Chain Organizations Have Built the Necessary Capabilities to Deliver on Future Performance," press release, February 18, 2025 (survey of 579 supply chain professionals, July–October 2024) — gartner.com
- McKinsey & Company, “The Transformative Power of Integrated Business Planning,” May 25, 2022 — mckinsey.com
- McKinsey & Company, " Pricing in Retail: Setting Strategy," April 1, 2015 — mckinsey.com
- Bain & Company, “Good Sales and Operations Planning Is No Longer Good Enough,” April 12, 2016 — bain.com
- Booper, internal product data (`context/socle_booper.md` §3) — GENIUS Predict, scenarios, and AI explanations.
Automating pricing doesn’t take decision-making away from humans; it eliminates the need for manual data entry. Two independent studies highlight the issue: AI-driven retailers see a 5 to 10% increase in gross margin (BCG), but 60% of AI projects not supported by AI-ready data will be abandoned by the end of 2026 (Gartner). The key difference between the two: a governance framework—including business rules, a designated owner, and a pilot category with a control group—established before automation, not after.
Not all AI systems are created equal when it comes to pricing. Statistical rules, predictive machine learning, and generative AI: these three technologies are often lumped together, even though they address different needs and inform different decisions.
Deciding and executing are two different things in AI pricing. Most reliable systems either carry out actions that have already been approved or make recommendations—they do not make decisions on their own in high-stakes cases.
The balance is struck by weighing the stakes and scope of each decision. Organizations that succeed in their AI projects are those that have established clear human oversight, not those with the most sophisticated model.