There’s a quiet dysfunction running through most supply chains, and it lives inside something every planner touches every single day.
The forecast.
Not the concept of forecasting — that remains essential. But the way the forecast is used: as the primary trigger for material replenishment. As the signal that tells your buyers what to order, when to order it, and in what quantity. As the assumed truth at the center of your entire planning architecture.
That assumption is costing companies far more than most senior leaders realize — and it sits at the heart of what Erik Bush, EVP of Internal Operations at Algo, has been challenging in The Trillion Dollar Forecast, Algo’s webinar series on the structural forces limiting supply chain performance.
The Accuracy Illusion
Forecasting accuracy is one of those metrics that tends to look better the further you zoom out.
At a category level? Ninety percent accuracy is achievable. Impressive, even. S&OP conversations get a comfortable number to point to. But Bush is direct about what happens when you get closer to where the decisions are actually made:
“You get it down to the item level, fifty percent, sixty percent might be accurate — because every week or every month depending on your planning horizon, you’ll see massive variations.”
This matters because MRP — the planning logic embedded in virtually every ERP system on the planet — doesn’t operate at the category level. It operates at the item level. When the signal is wrong — and it is wrong, constantly — the system faithfully executes the wrong plan at scale.
The Noise Problem
Here’s where it compounds.
Because the forecast changes — weekly, monthly, sometimes more frequently in organizations desperate to keep up with shifting demand — the plan changes with it. Wally Leisure, VP of Business Development at Algo, sees this play out in conversation after conversation with supply chain leaders:
“I’ve heard people say that they will do a new forecast load several times in a month to drive material in that same month. That’s their only mechanism to trigger buyers and planners. And as you can imagine on the receiving end of that — if you are a buyer or a planner — imagine how much you hate Mondays.”
Orders that were placed get cancelled. Suppliers who were told to hold get told to rush. This is the bullwhip effect in practice, and it doesn’t respect industry boundaries. As Bush points out, even environments that seem stable on the surface aren’t immune:
“Even in an automotive environment, if you’re a Tier 1 supplier feeding directly into an assembly plant, you would think that environment is relatively stable. We’ll demonstrate how week over week that signal is jumping up and down — because they don’t know what their customers are buying.”
The supply chain isn’t broken. It’s doing exactly what it was designed to do. The design is the problem.
Why We Keep Doing It
If the problems with forecast-driven replenishment are well understood — and they are, documented for decades — why does the practice persist?
Part of the answer is inertia baked into the systems themselves. Part of it is the seductive simplicity of the model. If we could just make the forecast better, the thinking goes, the rest would follow. So organizations invest in more sophisticated forecasting tools, better data inputs, AI-driven demand sensing. The forecast gets marginally better. The fundamental dependency doesn’t change. Results disappoint.
And part of it, as Leisure observed from inside industry, is that the stakeholders driving technology decisions often aren’t the ones feeling the pain:
“The finance groups were getting disappointed, frustrated with reporting or numbers, and the answer was, let’s get a new ERP system. Supply chain always is an afterthought.”
A Different Model
The alternative isn’t to abandon forecasting. Bush is clear on this:
“Does this mean we don’t need to do forecasting? Hell no. Forecasting and demand planning are essential activities. They are at the foundation of how you do S&OP — but the way we utilize forecasting could be changed.”
In a demand-driven model, strategic buffers are positioned at key points in the supply chain — sized based on variability, lead time, and consumption rates. Replenishment is triggered by actual demand flowing through those buffers, not by a predicted future state. The forecast informs the size and positioning of the buffers. It doesn’t drive day-to-day execution.
The effect is significant. Planners work from a signal that reflects reality. Suppliers get more stable, predictable orders. Service levels improve because you’re no longer chasing a moving target.
The Real Question for Senior Leaders
If you’re a supply chain executive, a CFO watching working capital metrics, or a CEO wondering why your inventory position never seems to improve despite ongoing investment in planning tools, this is the question worth sitting with:
How much of our planning architecture is built on an assumption of forecast accuracy that we know — at the item level — doesn’t hold?
The answer, in most organizations, is: most of it.
Bush frames the opportunity plainly:
“The stakes are way too high. A trillion dollars of inventory just in the U.S., just in manufacturing. If we could start to release that inventory, we could buy more production resources, address other fundamental challenges — using that working capital in a manner that’s much more conducive to where we’re going.”
The forecast will always have a role in supply chain planning. The question is whether it deserves the role it currently holds.
Want to go deeper? Bush and the Algo team explore these questions across a four-part webinar series — from diagnosing the forecast problem to designing a demand-driven operating model and managing the organizational change it requires.
