Forecasting as a commitment problem [Interview w. Gazi]
Issue #25 - How demand planners protect decision windows before uncertainty hardens into reality
I have implemented demand planning systems. I have chased forecast accuracy as if perfection were one algorithm upgrade away. Every technology vendor promises the same thing. Higher accuracy, fewer surprises, better decisions.
What those promises rarely acknowledge is what a forecast actually does. It quietly commits the organization to spending cash, building inventory, and locking production capacity long before demand is known. Once those commitments are made, optionality disappears.
My original belief did not survive my conversation with Gazi Sanaul Hasan, a Global Demand Planner. He does not see forecasting as a problem of accurate prediction. He sees it as a mechanism for governing how much risk an organization is willing to accept under uncertainty.
Yes, moving from Excel to advanced planning algorithms improves the baseline. A solid statistical model matters. But the gains flatten quickly. Once that foundation is in place, improvements become incremental. At that point, the limiting factor is no longer the model. It is judgment.
As Gazi made clear, forecasting does not fail because algorithms are too weak. It fails when numbers are treated as promises rather than as inputs to decisions that consume resources before reality has a chance to intervene.
Demand planner: the air traffic controller
When Gazi described his role, he did not talk about optimization, algorithms, or forecast accuracy. He said he sits in the air traffic control tower.
The metaphor matters.
An air traffic controller does not decide where planes want to go. They do not control the weather. And they do not promise that every flight will land on time. Their job is to manage congestion, sequencing, and risk within fixed constraints. They decide what can land safely, when, and at what cost if conditions deteriorate.
That is how Gazi sees demand planning.
At Molton Brown, a British fragrance company operating across the UK, Europe, the US, and an expanding Asian footprint, demand does not arrive as a single stream. It comes through multiple channels, each with different expectations and constraints. Amazon behaves differently from direct e-commerce. Retail stores want different pack sizes than hotels. A luxury hotel ordering thousands of 30 ml bottles follows a completely different rhythm than a retail channel selling 300 ml products.
The demand planner does not control these demands. What they control is how demand information is translated into commitments that consume cash, inventory, and production capacity.
That translation is where most organizations misunderstand forecasting.
What a forecast actually is
Forecasts are often treated as promises. A number is published, and the organization behaves as if it represents a future that will happen with sufficient certainty. When accuracy is high, the confidence increases. When accuracy is low, the response is to search for better models.
This framing is misleading.
Gazi is explicit that a forecast is not a prediction of demand and not a promise to the business. It is a decision support tool. Its purpose is to help different parts of the organization decide how much risk they are willing to carry.
A forecast immediately triggers real actions. Finance prepares budgets. Procurement orders raw materials and packaging. Production schedules labor and capacity. Cash is spent long before revenue materializes, if it materializes at all.
This also explains why Gazi shows little interest in unconstrained demand. As consultants, we often treat unconstrained demand as the starting point, a pure signal of what the market wants. But in an operating organization, that number has no decision attached to it. It does not tell finance how much cash to risk, supply how much capacity to lock, or planners how much inventory exposure is acceptable. Unconstrained demand is a question of possibility. Demand planning is a question of commitment. Once commitments are made, optionality disappears. From that perspective, unconstrained demand is not wrong. It is simply not important.
For that reason, demand does not exist in isolation. It must pass through three hard constraints, which he describes as runways:
• Cash
• Inventory
• Production capacity
These constraints determine what can safely land.
Cash is not a downstream finance consideration. Forecast quantities directly shape financial exposure. Every unit forecasted pulls cash forward. The organization is not just forecasting sales. It is deciding how much financial risk it is willing to absorb while waiting for demand to materialize.
Inventory is where forecasting mistakes become visible. Excess stock is not a neutral buffer. Once inventory exists, optionality is reduced. It creates holding cost, obsolescence risk, and downstream decisions that are difficult to reverse. For Gazi, inventory is not a shock absorber. It is a constraint that forces discipline.
Production capacity is the most rigid constraint. Forecasts cannot be freely adjusted in the short term without destabilizing upstream suppliers, labor planning, and production sequencing. This is why Gazi works with fixed forecast brackets and lead-time-driven planning horizons. Inside those brackets, change is deliberately limited.
Taken together, these constraints define the decision window within which forecasting operates. A forecast without its assumptions, reasoning, and limits is incomplete. Numbers alone are not useful. Context is what allows the organization to decide.
When errors turn into disruption
One of the most revealing moments in our conversation is a simple story. A number in the forecast is wrong. An extra zero is added. Or a supplier provides incorrect raw material data.
Nothing breaks.
The system continues to function exactly as designed. The forecast is published. Production starts. Procurement orders materials. Capacity and labor are committed. No one downstream stops to ask whether the number makes business sense.
By the time the error is discovered, the organization has already converted a data mistake into physical reality.
This is not a failure of individuals. Errors are inevitable when managing thousands of SKUs. What turns a mistake into a disruption is how the organization treats the forecast.
When forecasts are treated as instructions rather than hypotheses, small errors accumulate silently. There are no alarms. No dramatic collapse. Just a sequence of reasonable actions based on a flawed assumption.
This is slow-motion disruption.
The missing control is not better data. It is business sense.
Business sense as a control mechanism
Gazi is clear that forecasting cannot rely solely on systems or models. Every function involved must carry an intuitive baseline for what makes sense.
What does this SKU usually sell per month?
What would count as a believable growth rate?
What would be so far outside the norm that it requires explanation?
Without that shared sense, the organization becomes vulnerable to automation bias. Numbers are accepted because they exist, not because they are plausible.
Gazi emphasizes that forecasting is not “just an Excel file.” Without interpretation, Excel turns small errors into large commitments. Business sense acts as a distributed control mechanism, catching distortions before they harden into inventory, cash exposure, and capacity lock-in.
Outliers are decision problems, not forecasting problems
Outliers are inevitable in demand planning. The question is not how to eliminate them, but how to respond.
Gazi uses the concept of an acceptance rate as a believability boundary. A 20 percent uplift might be plausible without extraordinary justification. A 90 percent uplift is not. At that point, the number demands a narrative.
The planner’s job is to determine whether an outlier represents signal or noise. That means asking what changed and why now. Is it a promotion, a one-time buy, a channel shift, a launch, or simply a data error?
Only when the deviation exceeds the acceptance rate does it escalate into S&OP. Not to validate the number, but to force a decision. Sales may want coverage. Supply wants stability. Finance wants controlled exposure. The meeting exists to decide which risk the organization is willing to fund.
Outliers are not forecasting failures. They are moments where trade-offs must be made explicit.
Anticipation is not prediction
Statistical forecasting provides a baseline. But Gazi is clear that the real work begins when planners look beyond historical data and start monitoring weak signals that challenge underlying assumptions.
One example was the temporary closure of Amazon warehouses in the UK due to health issues. The event did not show up in sales data immediately, but it altered fulfillment conditions. Gazi did not assume demand collapsed. He judged it as a temporary headwind and deliberately cooled the forecast for the UK market.
The adjustment was not about being right. It was about avoiding overcommitment while uncertainty was elevated.
This is anticipation as assumption management. Responding to changes before they show up in actuals, not by chasing signals inside fixed decision windows, but by adjusting how much risk the organization carries.
The same logic applies to new product launches, where no historical baseline exists. Forecasts become hypotheses shaped by judgment and revisited as early signals emerge, rather than treated as fixed commitments.
Anticipation is less about seeing the future clearly and more about noticing when the present starts to shift.
The planner’s real job under uncertainty
Gazi is explicit in his advice to new planners. Uncertainty is permanent. Better tools and more data will not remove it. Treating uncertainty as a failure of planning leads to false confidence and overcommitment.
The planner’s value is not prediction accuracy. It is early signaling.
That means raising a hand when assumptions weaken, calling out risk before it appears in actuals, and making uncertainty visible rather than absorbing it quietly into a single number.
The planner does not decide risk alone. Their role is to frame the trade-offs clearly so the organization can choose.
What happens if demand is lower than forecast?
What happens if it is higher?
How do those outcomes affect cash, inventory, and capacity?
Forecasting, in this sense, is not about being right. It is about protecting decision windows before commitments become irreversible.
The demand planner is not there to promise the future. They are there to govern how much risk the organization accepts while moving toward it.
If you enjoyed this post and want to support my work:
Have you lived through a supply chain disruption? I am interviewing supply chain professionals for Supply Chain Shift. Please message me if you would like to share your story.
Sources
Interview with Gazi Sanaul Hasan, 22 January 2026