Supply chain planning is more sophisticated than ever. The plans it produces are more fragile than ever. It’s time to ask why.
Supply chain planning has never been more sophisticated – and never more misaligned with reality. We have faster systems, better models, and more data than ever before. And yet, planning teams are still firefighting every week.
That is not a tooling problem.
It is an objective function problem.
Arkieva’s CEO, Anand Iyer, made an important argument in his latest blog: speed without plan quality is a trap. Faster replanning just accelerates the cycle of failure. He’s right. But I want to push one level deeper – because the firefighting planners experience every week, the disruptions that blow up plans that seemed perfectly reasonable, the hours spent stabilizing rather than deciding – none of that is a technology failure. The system is working perfectly. It is just optimizing for the wrong reality.
The fundamental flaw in supply chain planning is architectural, not operational. We built systems to optimize efficiency in a world that required resilience
The Plan That Looks Perfect Until It Isn’t
Traditional supply chain planning follows a well-established sequence. Collect demand signals. Run a forecast. Optimize supply against that forecast – balancing cost, capacity, and service commitments. Publish a plan. Planners review exceptions. Execution begins.
It is a rational architecture. It was designed for a world where supply chains were relatively stable, disruptions were episodic, and the main challenge was computational – how to solve a large optimization problem efficiently.
That world no longer exists.
Today, supply chain volatility has become structural. Demand signals shift weekly. Supplier reliability fluctuates. Lead times drift. Geopolitical events ripple through networks in ways that compound rather than cancel. And in process industries – chemical, food, petrochemical, oil – these risks interact with physical constraints that make fragility nonlinear: a shelf-life limit, a campaign minimum, a yield variability that cascades through multi-stage production.
In this environment, traditional planning architecture has a critical blind spot: risk is discovered after plans are created.
What does this look like in practice:
Chemical manufacturers run their weekly planning cycle. The optimizer produces a plan that is cost-efficient and service optimal against the deterministic inputs it was given. Three days later, a supplier signals a delay. The plan is now structurally infeasible. Planners scramble. Expedite orders go out at multiples of normal cost. The S&OP team reconvenes to re-plan.
This is not an exception.
It is the operating model. Nothing in this scenario is surprising.
And that is exactly the problem.
The Wrong Objective Function
To understand why this keeps happening, it helps to think about what the planning system is actually solving for. At its core, a supply chain optimizer minimizes a cost function subject to a set of constraints. The objective is something like:
Minimize: Cost + Penalty for Service Shortfall (assuming the forecast is correct)
That last clause is almost never stated, but it is always implied.
And in today’s environment, it is almost always wrong.
This is a reasonable objective for a stable, predictable environment. The problem is that it is mathematically incomplete for a volatile one.
What is missing from that equation? Risk.
Risk as a first-class input that shapes what the optimizer is actually trying to achieve- not a dashboard once can check after the plan runs, not a scenario one can model in a spreadsheet after the fact.
A Useful Mental Model
A useful way to think about this is the difference between two types of plans:
- Cost-optimal plans: Plans that assume the world behaves as expected
- Reality-optimal plans: Plans that perform well when the world does not
Most planning systems are designed to produce the former. Most planners are forced to operate in the latter. The gap between them is where the firefighting live
The plan the system produces may be optimal against a deterministic forecast. But it is almost certainly not optimal when one accounts for the probability that the forecast is wrong, that suppliers will be late, that yields will drift, or that demand will arrive in unexpected patterns.
In other words:
The planning system is solving the wrong problem.
It is finding the optimal answer to a simplified version of reality – and handing that answer to planners who then spend their time managing the gap between the plan and the real world.
Why Visibility and Speed Miss the Point
Anand covered the speed trap well. But speed is only one of two dominant responses the industry has produced. The other is visibility: build a digital twin, model the network, alert planners when something goes wrong. Both are genuinely valuable. And both share the same fundamental flaw.
Visibility tells planners what broke after risk has materialized. Speed helps recover faster once the plan is already failing. Both approaches treat risk as something to be detected and responded to. Neither embeds risk into the planning decision itself. They are downstream responses to an upstream architectural problem.
And that upstream problem is this: the objective function the planning system is minimizing was never designed to account for risk at all.
We have more data, speed, and visibility than ever. We are still optimizing for a simplified version of reality and calling it a plan.
This distinction matters enormously in practice. A planner who is told, after the fact, that a supplier is at risk still has to manually assess the impact, run scenarios, and make judgment calls under time pressure. A planner whose system incorporated that supplier’s reliability into the original plan starts from a fundamentally different position: a plan designed with that risk already considered.
Shift-Left Risk: Designing for Reality
In software engineering, Shift-Left Security did not emerge from a single insight. It emerged from three converging realities – and understanding all three is what makes the supply chain parallel so precise.
First: cost of fix scales with time. A vulnerability caught in design costs a fraction of one discovered in production – in dollars, time, and reputational damage. So Software Development moved the check left. In supply chain: a fragile assumption caught during planning costs a rounding error compared to one discovered mid-execution, when expedites are flying, commitments are at risk, and the S&OP team is convening an emergency session. The economics of moving risk detection upstream are just as compelling.
Second: the external threat landscape changed, making reactive defense untenable. Software security did not shift left because engineers suddenly got more disciplined. It shifted left because the external threat environment – state-sponsored hackers, third-party component vulnerabilities, zero-day exploits – escalated to a frequency and sophistication that reactive patching simply could not keep pace with. The same structural shift has happened in supply chain. Geopolitical realignment, tariff volatility, climate-driven disruptions, supplier concentration risk – these are not episodic shocks anymore. They are the operating environment. Planning architecture designed for a stable world and patched reactively for a volatile one is not a strategy. It is a liability.
Third: robustness cannot be added at the end – it must be foundational. Secure software is not software with a security layer bolted on at deployment. It is software where security was a design principle from requirements through release. You cannot audit your way to a secure system after the architecture is set. The same is true of resilient supply chains. You cannot dashboard your way to resilience after a fragile plan has been published. Risk must be a foundational tenant of the planning process itself – from the demand signal through production scheduling and execution. It should be a first-class variable embedded in how the plan is built, present from the demand signal through execution.
None of these three realities are different for supply chain risk. Which is why the response cannot be different either. Supply chain planning needs the same architectural shift that software engineering made a decade ago. Call it Shift-Left Risk.
Instead of the traditional sequence:
Forecast → Optimize → Add Risk Controls
The Shift-Left Risk architecture moves toward:
Risk Signals → Risk-Adjusted Forecast → Risk-Constrained Optimization → Execution
Risk is no longer something that happens to a plan. It is embedded in how the plan is built.
This means supplier fragility scores influence sourcing allocations before the optimizer runs. It means demand uncertainty is expressed as a distribution, e.g., a P10, P50, and P90 rather than a single point estimate that will almost certainly be wrong. It means feasibility constraints like campaign logic, shelf-life limits, allergen changeovers, minimum run lengths are embedded in mid-term planning layers, not discovered when detailed scheduling tries to execute an impossible plan.
And critically, it means the objective function evolves:
Minimize: Cost + Risk Exposure Penalty − Stability Credit
Plans are now evaluated not only on what they cost if everything goes right, but on how they perform when things go wrong – which, in today’s environment, they reliably do.
What This Means for Planners, Leaders, and Business
For planners, the experience is different in ways that matter immediately. Instead of starting each week with a plan that assumes everything will go according to forecast, they start with a plan that has already accounted for the most likely failure modes. Their job shifts from firefighting to decision-making, from repairing plans to choosing between risk-informed alternatives.
For supply chain leaders, the strategic implication is significant. Plan stability becomes a measurable metric – not just a feeling. The question stops being “why did we replan three times this week?” and starts being “what is our shock absorption capacity relative to our risk exposure?” That is a conversation a planner can have with a CFO. That is a metric one can track over time.
For the business, the value shows up where it counts: margin protection, working capital efficiency, and service reliability in conditions where competitors are still firefighting. Resilience, built by design rather than recovered through heroic effort.
The Planning Paradigm Is Shifting
Every major shift in supply chain planning has followed the same pattern. What initially feels like an enhancement eventually becomes the baseline.
Risk-native planning will follow the same path.
The only real question is timing.
In five years, planning systems that ignore risk as a first-class variable will look as incomplete as systems that once ignored capacity or lead times.
Until then, many organizations will continue producing plans that look elegant on Monday and need repairing by Wednesday.
Risk is the environment now. Planning systems must be built for it.
Risk is not an afterthought. Risk is a first-class planning variable.
