TL;DR
Production planning determines what to make, in what quantity, and when, so that demand is met without excess inventory or idle capacity. It runs on a hierarchy from sales and operations planning down to detailed scheduling, with MRP calculating material requirements in the middle. The recurring failure is not the calculation. It is that the plan assumes its inputs are current, and between runs they quietly stop being current.
Key Takeaways: Production planning spans S&OP, the master production schedule, MRP, and shop floor scheduling, each operating at a different horizon. MRP calculates from three inputs: the bill of materials, inventory and supply position, and demand. It assumes open supply is accurate and lead times are fixed. In practice supplier dates move, quantities change, and shipments split, and those updates arrive as messages rather than data.
What is production planning?
Production planning is the process of determining what a business will produce, in what quantities, using which resources, and on what timeline, so that customer demand is met without accumulating excess inventory or leaving capacity idle.
It is a balancing exercise between three pressures that pull against each other. Service level, meaning delivering what customers ordered when they expect it. Inventory, since holding material is expensive and holding the wrong material is worse. And capacity utilization, since machines and labor represent fixed cost that idles when the plan is wrong.
Most manufacturers do not have a single production plan. They have a hierarchy of them operating at different horizons and levels of detail, and the distinctions matter because failures at one level present as symptoms at another.
The planning hierarchy
Sales and operations planning operates at the longest horizon, typically months to a year and a half, working at product family rather than item level. It reconciles the demand plan with available capacity and is a management process as much as a planning one.
The master production schedule translates that into specific finished items in specific periods. It sits between the aggregate plan and material calculation, and it is what drives MRP.
Material requirements planning calculates what components and materials are needed, in what quantities, and by when, to support the master schedule. It generates planned purchase orders and work orders.
Capacity requirements planning checks whether the resulting plan is actually executable against available machine and labor capacity, since MRP by itself assumes infinite capacity.
Production scheduling operates at the shortest horizon, sequencing specific jobs on specific resources, often day by day or shift by shift.
A plan can be sound at the S&OP level and undeliverable at the scheduling level, which is why shortages surface on the shop floor rather than in the planning meeting.
What MRP actually calculates from
MRP is the engine most of this rests on, and it takes three inputs.
The bill of materials, defining every component required to build the finished item, including multi-level structures where subassemblies have their own BOMs.
Inventory and supply position, covering current stock, allocations, and open supply, meaning purchase orders and work orders already released but not yet received.
Demand, from firm customer orders, forecasts, or both.
From these it calculates requirements by period and nets them against supply, producing planned orders and exception messages. The logic is sound and well established.
It also carries two assumptions that are rarely stated explicitly: that lead times are fixed and that open supply is current. Both are reasonable simplifications. Neither survives contact with a supply base.
Why plans decay between runs
Here is the failure mode that explains most planning frustration, and it is not a configuration problem.
When a buyer releases a purchase order, the work moves outside the planning system and into supplier execution. From that moment, the ERP holds a record of what was ordered and what date was expected at the time of ordering. Reality then proceeds independently.
Dates move. A supplier confirms a different week than the one requested. Quantities change, and a full order becomes two partial shipments. Lead times shift, sometimes by weeks and often with little notice, and lead time variability is consistently cited as the single largest planning challenge by manufacturers. Orders go on hold over a pricing discrepancy. A supplier confirms one line on a multi-line order and queries another.
Each of those is a change to open supply, which is one of the three inputs MRP depends on. If the change does not reach the ERP promptly, the next MRP run executes correctly against information that is no longer true.
The result is a plan that looks organized and is not accurate. This is why manufacturers experience shortages and line stoppages while the system still shows material as available. The calculation was right. The input was stale.
The commitment signals that keep a plan true
Keeping open supply current means capturing a specific set of signals against each purchase order line:
Acknowledgement, confirming the supplier has received and accepted the order at all.
Commit date, confirming whether the date in the ERP is still the date the supplier expects to meet.
Quantity confirmation, establishing whether the full order will arrive or whether planners should expect a split.
Lead time change, flagging when a supplier’s actual lead time has moved away from the value held in the item master.
Any one of these being out of date undermines the plan. All four are properties of a specific PO line rather than a supplier in general, which is why supplier scorecards do not substitute for them.
The other decay: master data
Alongside supplier execution, planning quality erodes through the reference data the calculation reads.
Common failures are inaccurate lead times in the item master, outdated bills of material that no longer reflect what engineering actually released, wrong units of measure, missing lot sizes and order minimums, and inconsistent item status codes. Engineering changes create mismatches between what was designed, what was approved, and what is physically stocked. This is the same failure pattern that master data management programs exist to address.
None of these announce themselves. They surface as a shortage, an over-order, or a plan that nobody trusts, which is the most expensive outcome because it drives planners back to spreadsheets alongside the system.
Where the work actually goes
Look at how planning and procurement teams keep the picture current, and the character of the work is clear.
Supplier confirmations arrive by email, as PDF acknowledgements, as portal notifications, or in the body of a message. Someone reads them, works out which PO line each refers to, determines whether the date, quantity, or price differs from what the ERP holds, and updates the record. When a supplier confirms one line and queries another, someone splits the handling. When a shipment is partial, someone adjusts the expected receipt.
This is reading and cross-referencing rather than planning judgment, and it scales with the number of PO lines and the volatility of the supply base rather than with the size of the planning team. When the volume of change exceeds the capacity to process it, updates lag, and the lag is exactly the gap between the plan and reality.
Teams compensate through effort: inbox monitoring, shared spreadsheets, expediting calls, and the standing morning meeting to work out what actually arrived. Those are symptoms of an input problem being absorbed by people.
Where automation fits
The useful conclusion from this is that improving planning accuracy often has less to do with the planning system than with the flow of information into it. Replacing an MRP engine does not help if the open supply it reads is three weeks behind.
Automation that can read unstructured messages and reason about their contents addresses the input layer directly: interpreting a supplier acknowledgement whatever form it arrives in, matching it to the correct purchase order line, identifying whether the confirmed date, quantity, or price differs from the ERP record, updating the record, and escalating the cases that genuinely need a buyer’s judgment rather than a data entry decision.
The effect is on latency between a supplier saying something and the planning system knowing it, which is the variable that determines whether the next MRP run is calculating against reality.
Because these updates drive purchasing and production commitments, each one needs to be traceable. A confirmation applied incorrectly propagates through the plan, and a date changed by a process nobody can inspect is difficult to unwind when a shortage appears three weeks later.
To be clear about scope, Kognitos is not a planning system. It does not run MRP, generate a master production schedule, or sequence jobs, and your ERP or APS platform remains the right system for those. What it addresses is the document and message traffic that keeps the planning inputs current, working alongside those systems and recording what was updated and why.
For related processes, see our guides on master data management, purchase order automation, procurement automation, AI in supply chain automation, and AI automation in manufacturing. To see how deterministic AI keeps supplier commitments current in your ERP, book a demo or try the platform.
Getting started
Two diagnostics worth running before changing any planning software.
Measure the age of your open supply data. Take a sample of open purchase order lines and check how many carry a confirmed date that the supplier has actually restated recently, versus a date that has sat unchanged since the order was raised. That proportion is the real accuracy of one of MRP’s three inputs.
Then look at where planners spend their time. If a meaningful share goes to establishing what is actually arriving rather than deciding what to build, the constraint sits in the input layer, and no planning engine will resolve it.
