The decoupling point is the position in a manufacturing flow where you stop building against a forecast and start building against a confirmed customer order. Everything upstream of that point is replenished to stock and driven by demand history, while everything downstream is triggered by an actual order.
Getting its location right is one of the highest-leverage decisions in operations, because it sets your lead time, your inventory investment, and how much variability you can absorb without either starving customers or drowning in work in process.
This guide explains how to place the customer order decoupling point and match your pull strategy to demand variability.
The decoupling point separates two very different control logics. Upstream, you run a replenishment pull system: parts and subassemblies are made to refill a buffer, and consumption from that buffer signals the next build. Downstream, you run an order-driven pull: nothing moves until a customer commitment exists. The buffer that sits at the decoupling point is what lets these two logics coexist without one destabilizing the other.
This position is often called the customer order decoupling point (CODP), and it maps directly to the classic strategy archetypes:
The further downstream you place the point, the shorter your customer-facing lead time but the more finished inventory risk you carry. The further upstream, the leaner your inventory but the longer the wait.
Placement balances two competing pressures: the lead time the customer will tolerate versus the variability you face in demand and product mix. When customer-tolerated lead time is shorter than your cumulative production lead time, you are forced to hold a buffer somewhere upstream so part of the work is already done before the order arrives. The decoupling point is where you choose to hold that buffer.
A useful principle: push commonality upstream and variety downstream. Standard, high-volume, low-variability components belong in a make-to-stock zone before the decoupling point, because forecasting them is accurate and cheap. Customer-specific variety belongs after the point, where it is pulled by real orders.
This is the logic behind product platform and postponement strategies, and it is why the decoupling point often lands exactly where a product's bill of materials fans out from few common parts into many variants.
Consider a company building industrial control panels. The routing has four stages with these lead times:
Cumulative production lead time is 5 + 6 + 3 + 2 = 16 days. Customers, however, will only wait 6 days from order to delivery. If you run pure make-to-order from raw material, you miss the promise by 10 days on every order.
Now place the decoupling point after Stage 2. Stages 1 and 2 run make-to-stock, replenishing a buffer of wired backplanes. When an order lands, only Stages 3 and 4 remain: 3 + 2 = 5 days, comfortably inside the 6-day window.
The customer-specific variety (Stage 3) stays downstream of the point, so you never forecast a specific configuration. You only forecast demand for the generic backplane, which is far more stable.
How big should that buffer be? Suppose average demand is 20 backplanes per day, the replenishment lead time to refill the buffer is 11 days (Stages 1 and 2), and daily demand has a standard deviation of 6 units.
A standard cycle-plus-safety approach gives a buffer target roughly equal to demand over the lead time (20 x 11 = 220 units) plus safety stock. Using a service factor of 1.65 for about 95 percent service, safety stock is about 1.65 x 6 x the square root of 11, or roughly 33 units.
A buffer target near 250 units of generic backplanes lets you hit a 6-day promise on almost every order while forecasting only one stable item instead of dozens of variants.
Demand variability decides which pull mechanism sits at and around the decoupling point. High-volume, stable items are ideal candidates for a fixed-quantity kanban replenishment loop: consumption pulls a card, the card authorizes a refill, and the buffer self-regulates. Lumpy or seasonal items are poorly served by fixed kanban; they need a reorder-driven approach where the trigger reflects the real signal.
Two levers matter here. First, set the trigger correctly using a disciplined reorder point that accounts for lead time and demand during that lead time. Second, size the protective buffer with an honest safety stock calculation rather than a round-number guess.
Segmenting your catalog with ABC analysis keeps you from applying the same policy to a runner and a stranger: A-items in a tight pull loop, C-items on simpler min-max rules.
The overall discipline is a pull system: nothing is produced until a downstream signal authorizes it. The decoupling point is simply the boundary where the signal changes from an internal replenishment trigger to an external customer order.
A decoupling point only works if the upstream stages can actually refill the buffer on schedule. If the fabrication and backplane cells break down unpredictably, the buffer empties, and your short downstream lead time evaporates the moment stock runs out. This is where flow control meets asset reliability.
The upstream replenishment zone usually contains the constraint that governs throughput, so it deserves Theory of Constraints thinking: protect the bottleneck, and schedule the whole zone to keep the decoupling buffer full. A drum-buffer-rope schedule does exactly that, pacing releases to the constraint drum.
Meanwhile the equipment feeding the buffer needs stable uptime, which is why overall equipment effectiveness and a solid preventive maintenance program are not side issues but preconditions for the whole make-to-stock zone to hold. Watching inventory turnover around the point tells you whether the buffer is sized right or quietly bloating.
Placing a decoupling point is a data problem before it is a design problem. You need to know your real cycle times per stage, where uptime losses actually occur, and whether the upstream cells can keep the buffer full.
Fabrico gives you that real-time foundation: live OEE and production monitoring show true throughput and downtime at each stage, including on older machines with no PLC, through computer vision. That data lets you set the decoupling point on measured lead times rather than assumptions.
On the reliability side, Fabrico is a field-ready CMMS with work orders, asset records, preventive scheduling, and spare-parts tracking, so the upstream cells that replenish your buffer stay dependable. You can explore the monitoring side in the OEE solution overview and the maintenance side in the CMMS solution overview. Fabrico is EU-built with EU data residency, so the operational data behind your pull system stays under European governance.
They answer different questions. The bottleneck is the resource that limits total throughput and can sit anywhere in the routing. The decoupling point is the control boundary where production switches from forecast-driven to order-driven. They can coincide, and often the constraint sits in the upstream make-to-stock zone, but they are separate concepts.
You size a buffer at the decoupling point to absorb demand variability, and you protect the bottleneck to defend throughput.
Yes. Multi-product plants routinely run different strategies per product family. A high-volume line may keep its point at finished goods (make-to-stock) while a configurable line keeps its point at subassembly (assemble-to-order) and a bespoke line runs engineer-to-order.
Even within one product you can hold a strategic buffer at both a component level and a subassembly level. The goal is always the same: match each product's placement to its own demand variability and tolerated lead time.
Revisit whenever the inputs move. If customer-tolerated lead time shrinks, product variety grows, or demand variability changes, the optimal position shifts. A practical cadence is a quarterly review against measured cycle times and service levels, plus a trigger-based review any time you launch a product platform, add significant configuration options, or see buffer turns drift outside their target band.
Want to place your decoupling point on measured reality instead of guesswork? See how Fabrico's real-time OEE and CMMS give you the stage-level data and uptime to make the call. Book a Fabrico demo and start building your pull system on a solid data foundation.
Login
Blog
