Keystone at Zero100: From Forecasts to Decisions at Scale

At the Zero100 event in October, Keystone joined global supply chain and operations leaders to address one of the most persistent challenges facing enterprises today: how to make better decisions under uncertainty. In a session led by Greg Richards, Keystone’s leadership team laid out why traditional forecasting approaches are no longer sufficient—and what it takes to move from static plans to decision intelligence at scale.

Forecasting Is Not the End Goal—Decisions Are

Greg opened the session with a simple but provocative assertion: forecasting has become disconnected from the decisions it is meant to support. Despite decades of investment in planning systems, most organizations still rely on point forecasts, manual buffers, and heuristics to make high-stakes decisions around inventory, production, service levels, and capital deployment.

The result is a familiar pattern: excess inventory alongside poor service, frequent replanning cycles, and teams spending more time reconciling numbers than acting on them.

Keystone’s view is that forecasting should not be judged by statistical accuracy alone, but by how well it enables real operational and financial decisions.

The Core Problem: Point Forecasts and Heuristics

A central theme of the presentation was the structural limitation of point forecasts. Greg explained that most enterprise planning systems—whether embedded in ERP, APS, or S&OP tools—produce single-number forecasts and then rely on downstream rules to account for uncertainty.

This approach forces planners to compensate manually:

• Safety stock becomes a proxy for risk.
• Reorder points are set infrequently and rarely revisited.
• Production and capacity decisions are based on averages rather than distributions.

According to Keystone, this is not a tooling problem—it is a modeling problem. Decisions such as how much to produce, where to hold inventory, or what service level to promise are inherently probabilistic. Treating them as deterministic leads to systematic inefficiencies.

A Shift to Probabilistic, Decision-Centric Forecasting

Greg outlined Keystone’s alternative: Foundation Forecasting, a system designed explicitly to support decisions under uncertainty.

Rather than producing a single forecast, Foundation Forecasting generates probabilistic, multi-horizon forecasts across multiple levels of the enterprise—from market and product lines down to SKU, customer, and ship-to. These forecasts quantify uncertainty directly, producing distributions and quantiles that can be optimized against real economic tradeoffs.

This shift enables organizations to answer questions that traditional systems cannot:

• What inventory level minimizes total cost given asymmetric stockout and overstock penalties?
• How should production frequency change if demand uncertainty narrows or widens?
• What is the economically optimal promise date, not just the earliest possible one?

From Planning to Optimization and Automation

A key message from the session was that better forecasts alone are not enough. Value is unlocked when forecasts are directly connected to decision optimization and automation.

Greg described how Keystone’s platform integrates forecasting with optimization techniques from operations research to drive decisions across:

• Production planning and run frequency
• Inventory positioning across suppliers, factories, and distribution centers
• Replenishment policies such as safety stock, reorder points, and lot sizes
• Order promising (ATP/CTP) under uncertainty
• Commercial decisions including promotions, pricing, and clearance

Importantly, Keystone positions this capability as complementary to existing ERP and SAP environments, not a replacement. The platform improves the intelligence feeding planning and execution systems, while allowing organizations to retain their systems of record.

Why This Matters Now

The session closed by addressing why this shift is urgent. Volatility in demand, supply constraints, labor rigidity, and capital pressure have exposed the limits of traditional planning approaches. Enterprises can no longer afford slow planning cycles or inventory-heavy buffers as their primary risk management strategy.

According to Greg, organizations that succeed will be those that:

• Treat uncertainty as a first-class input, not an afterthought
• Design forecasts around decisions, not reports
• Separate intelligence from execution systems
• Move from manual planning to scalable, AI-enabled decision intelligence

A Clear Takeaway

Keystone’s message at Zero100 was clear: the next frontier in supply chain and operations is not better dashboards or faster planning runs, but better decisions—made consistently, economically, and at scale.

As Greg summarized, “Forecasts are the entry point. Decisions are the outcome. The companies that connect the two will define the next decade of operational performance.”

‍