When the signal arrives too late
At 7:10 on a Monday morning, a Tier-2 stamping supplier may have a kanban request for this week, an MRP proposal for next week, and a spreadsheet from a Tier-1 customer warning that the vehicle program could accelerate. None of those signals answers the question the production planner actually faces: how many pieces should the press run today to protect the assembly line without filling the warehouse with the wrong parts?
This is the gap that conventional demand planning often leaves at Tier-2 and Tier-3. A supplier receives a purchase order, a blanket order, a delivery schedule, or a manually adjusted forecast. The information may describe what the customer has already committed to buy, but not the pattern that is forming at the assembly plant. By the time a change is visible in the local ERP, the supplier may already be expediting steel, changing a die schedule, or asking operators to work around a shortage.
The result is a familiar cycle. A forecast misses a model changeover or a seasonal shift. The supplier adds safety stock. Demand then softens, leaving excess material and finished goods. The next shortage triggers an emergency reorder, a premium freight request, or an andon call. Each response is understandable in isolation. Together, they create nervousness in the supply chain.
The planning tools used by Tier-1 suppliers were not designed for the operating reality of every smaller supplier below them. They often assume cleaner data, longer planning horizons, and more control over upstream and downstream events than a Tier-2 or Tier-3 actually has. A new approach needs to begin with the suppliers that conventional planning has routinely underserved.
Planning from the line backward
A vehicle assembly plant does not consume a forecast in the abstract. It consumes parts at a takt time, according to a build sequence and a pull schedule. That schedule changes the requirements for the suppliers attached to it. A Tier-1 supplier may translate the plant schedule into an order for an assembly. A Tier-2 supplier then sees a requirement for a stamped bracket, machined housing, molded component, or electronic subassembly. A Tier-3 supplier may see only a material or process requirement several steps removed from the original signal.
At each handoff, timing and context can be lost. Order quantities can be rounded. Updates can arrive in different file formats or on different cadences. A planner may know that this week's release increased, but not whether the increase reflects a temporary schedule adjustment, a genuine ramp, or a model changeover. The BOM shows how parts relate to one another, but it does not by itself explain how demand will move through the network.
Lean Path starts with the assembly plant pull schedule and forecasts demand tier by tier. Its time-series model is trained on pull schedule history, seasonality, and model changeovers. The purpose is not to replace the planner's knowledge of press capacity, supplier lead times, or quality controls. It is to provide a clearer view of the signal before it becomes an urgent purchase order.
That distinction matters for a smaller operation. A planner at a Tier-2 company may be responsible for customer releases, production sequencing, raw material orders, inventory reviews, and delivery calls in the same day. A forecast that requires a large data project or a new planning department will not fit that environment. The useful question is more practical: can the supplier see a likely requirement early enough to adjust a run, a material order, or a kanban setting?
Why tier-by-tier forecasting changes decisions
Consider a plausible scenario involving a Tier-2 stamping supplier in Aichi. The supplier makes several brackets for a vehicle program and normally works with an 8 to 14 day material lead time. Its ERP contains customer orders and production history, while the Tier-1 sends weekly schedule updates. A model changeover causes the assembly plant's mix to shift over several weeks. The total number of vehicles is relatively stable, but the bracket mix changes faster than the supplier's safety stock rules.
In the first week, the planner sees a modest increase in one part number and adds a purchase order for sheet material. In the second week, a different part begins to rise, but the signal is obscured by order batching. By the time the pattern is clear, the press schedule has already been fixed and the supplier is considering overtime. Meanwhile, material purchased for the first part is moving more slowly than expected.
A tier-by-tier forecast would examine how the pull schedule propagates through the relevant BOM relationships and supplier tiers. It would not promise perfect visibility. It would give the planner a directional view of which requirements are strengthening, which are temporary, and which deserve review before an automatic reorder is released.
This is where the distinction between forecast and reaction becomes important. A forecast does not eliminate kanban, MRP, or ERP. It can sit alongside those systems and help a planner decide when their recommendations need investigation. The planner may still override a proposed order because of a die-maintenance plan, a PFMEA action, a quality hold, a minimum lot size, or a known constraint at a sub-tier supplier. The aim is better timing for those decisions, not blind automation.
Lean Path typically produces its first forecast in the third week of onboarding. Its pilot is an eight-week structured engagement covering up to 15 SKUs. This limited scope is deliberate. It allows a design-partner team in the Japan automotive sector to compare the forecast with the decisions already being made in daily planning meetings, rather than treating a dashboard as proof of value by itself.
What this approach can and cannot do
Based on design-partner pilot data, Lean Path has observed approximately 35% fewer unplanned reorders. That figure is qualified by the pilot setting, the participating SKUs, and the way each team managed its planning process. It is not a universal performance guarantee, and it should not be read as a promise that every supplier will achieve the same result.
There are also clear limits. A forecasting model cannot repair a missing or incorrect BOM. It cannot create capacity when a press, mold, heat-treatment line, or skilled operator is already fully booked. It cannot infer a customer decision that has not entered any usable schedule or history. It cannot replace APQP, process controls, traceability, or the quality judgment required when a part is out of specification.
Forecast quality also depends on the signal. If pull schedules are frequently overwritten without explanation, if part-number mappings change without governance, or if model-change information arrives after production has started, the forecast will have less to work with. The right response is not to hide that uncertainty. It is to show planners where the signal is weak and where a human review is required.
Our boundary is straightforward: Lean Path is not claiming to solve every automotive supply chain problem, and this article is not claiming that AI should replace MRP, ERP, kanban controls, or experienced planners. The founding premise is narrower. Tier-2 and Tier-3 suppliers need demand information that reflects how the assembly plant's pull moves through their tier, in a form they can use before shortage or excess becomes visible in the warehouse.
A practical starting point for underserved suppliers
For a supplier considering this problem, the first step is not a company-wide system replacement. It is to select a small group of parts where the cost of nervous planning is visible. Useful candidates may include SKUs with frequent expedites, repeated safety-stock adjustments, unstable order patterns, or a known model changeover. The planning team can then compare the incoming pull signal with actual production, material ordering, inventory, and reorder decisions.
- Map the selected SKUs to the relevant BOM and supplier tier.
- Review the history of pull schedules, releases, changeovers, and actual consumption.
- Separate normal seasonality from one-off schedule changes and known disruptions.
- Record when a reorder was triggered, why it was triggered, and whether it became necessary.
- Use the forecast as a planning input, with explicit review by the people responsible for capacity and quality.
This is also why Lean Path is being built as a small, customer-facing company. Takeshi Yoshida, CEO & Co-Founder, Kenji Tanaka, CTO & Co-Founder, and Rika Hayashi, Head of Operations, work directly with design partners rather than assuming that a generic planning workflow fits every supplier. The team is bootstrapped, and the early product is intentionally focused on Japanese automotive Tier-2 and Tier-3 supply chains.
Lean Path keeps automotive supply chains stocked without overbuying. Our AI forecasts demand across each tier of parts suppliers, so a plant orders what the line will need, not what it panicked over last week.
For Tier-2 and Tier-3 suppliers thinking about the difference between a customer release and the demand behind it, this is the beginning of a conversation. Contact Lean Path at [email protected] to discuss an early design-partner engagement and the parts where better visibility could change the next planning decision.