Why Delivery Promises Keep Missing
If sales says "we can ship by next Friday" and the order actually leaves the following Wednesday, again and again, the cause is probably not a careless planner. It is the way the date is calculated.
ERP MRP assumes infinite capacity. It gives each item a fixed lead time (say, five days) and counts backward from the due date. It never checks how loaded the machines already are that week.
Take a single press running two 8-hour shifts, five days a week: 80 available hours. Confirmed jobs already fill 70 of them. A new order for 1,000 pieces (1.2 minutes each, plus a 2-hour setup) needs 22 hours. Only 10 hours remain, so 12 hours of work spills into the following week. MRP cannot see the overflow and still answers "this week."
The gap is usually bridged by one planner's spreadsheet and experience. When constraints such as "we only have one die set" or "this part runs on only one machine" live in a single person's head, the quality of every delivery promise depends on that person being at their desk.
What APS Actually Does: Finite-Capacity Planning and CTP
APS (Advanced Planning and Scheduling) plans against finite capacity. Its defining feature is that it considers three things at once:
This changes how dates are promised. ATP (Available to Promise) answers from current stock and the uncommitted portion of production already planned. CTP (Capable to Promise) goes further: when there is no stock, it calculates when the order could be finished with the remaining capacity and materials.
With CTP in place, a sales rep enters the item and quantity, the system tentatively slots the order into the live schedule, and the rep replies on the spot. The half-day wait for production control to call back disappears.
Master Data to Fix Before You Start
An APS result can never be more accurate than the master data behind it. Check at least these four before implementation:
If an item is registered at 1.0 minute but really runs at 1.2 minutes, a plan built for 80 hours takes 96 hours on the floor. The setup matrix matters just as much. Where a light-to-dark changeover takes 15 minutes and dark-to-light takes 60, four changeovers a day cost either 60 or 240 minutes depending on the sequence.
Once the schedule drifts by even a couple of days, the shop floor goes back to spreadsheets and whiteboards. Lost trust is hard to win back with a better algorithm, so treat master data cleanup as the implementation itself, not as a preliminary step.
Dividing the Work Among ERP, APS, and MES
The three systems have distinct roles:
What matters is that the flow forms a closed loop. If MES actuals never return to APS, the plan stays frozen at Monday morning's assumptions and by Wednesday no longer describes reality.
Decide three rules at design time:
Without a frozen zone, every reschedule overturns material staging and die changes, and the floor ends up with more confusion than before.
Package or Custom Build?
The planning logic you need depends on the type of process:
Whichever route you choose, avoid rolling out to every operation at once. In most plants, one or two bottleneck operations determine the delivery date. Clean up master data for one of them, run a finite-capacity schedule there, measure schedule adherence, and only then extend upstream and downstream. That keeps the cost of failure small.
Advancing Production Planning with POLYGLOTSOFT
POLYGLOTSOFT builds production planning systems that derive real cycle and setup times from MES actuals and feed them into the planning logic. We support a phased approach: start with a scheduler for one bottleneck operation, then extend to CTP date lookup, ERP integration, and automated rescheduling.
With our subscription development service, you can add capabilities month by month without a large upfront investment. If you are ready to stop promising delivery dates from a spreadsheet, contact POLYGLOTSOFT.