Lesson 7: Reliable due dates

The last piece of the TOC way of production is the promise to the customer. A plant that delivers reliably can sell that reliability. One that can’t is stuck competing on price.

Why the classic planning systems disappoint

MRP has a compelling logic: with a feasible master schedule, a clean bill of materials, routings and inventory data, it can work out exactly when every part should be made or bought. If every resource followed the plan, inventory would be low and delivery on time.

Its weak spot is the fixed lead time. The time an order takes depends on how much other work is ahead of it in the queue, and MRP ignores capacity. The plan says a part needs three days; in practice the constraint is already booked for six. Overloads pile up in execution with no warning in the plan. The “closed loop” capacity check that followed in MRP II never really closed: fixing an overload in one place created another elsewhere, so plants accepted the overloads and replanned at month-end.

Advanced planning systems (APS, ASCP) promised to solve it mathematically: optimise against many constraints at once, commit dates against finite capacity, and reschedule instantly when something changes. In practice:

They try to optimise many constraints. Most plants seem to have many constraints only because there is too much work on the floor. Optimising all of them keeps the inventory up.
They institutionalise rescheduling. If every disruption triggers a new plan, the original promise to the customer was never really a promise.
Safety added to make it work isn’t optimal any more, and without feedback from execution it is either too much or too little.
Time-based priorities mislead in make-to-stock. An order may be “late” against its schedule while the warehouse is full of that item, and another “on time” while its stock is about to run out.

The underlying issue is the conflict from lesson 3. Both systems tried to plan their way out of it, with more data and more computing, instead of breaking it.

Pause and think: a plant reschedules every time a machine breaks down, so the plan always reflects reality. What happens to on-time delivery?

It suffers: every reschedule moves some orders’ dates, and disruptions spread through the whole plant
It reaches 100%
It is unaffected
It improves only for big customers

Rescheduling transfers each local disruption to the whole organisation, and the customer gets a new date instead of the promised one. Instead of adjusting the schedule for uncertainty, TOC protects it, with time or stock buffers.

The TOC answer: a good-enough plan and perfect execution

One constraint, not many. Choke the release and the apparent multiple constraints collapse into one. Now you need accurate data in only one place.
Buffers instead of rescheduling. Protect promises with time (or stock, for make-to-stock items) rather than replanning at every disruption.
Execution signals instead of imposed schedules. Work centres follow buffer colours, which tells them what is best for flow, and the reasons for reds show where to improve.
Daily, not monthly, planning. Orders are committed every day against the actual load, so there is no month-end skew to recover from.

Quoting a date you can keep

To promise a new order, look at the planned load on the constraint: how many days of work are already committed ahead of it. A safe date is roughly:

today + days of work already booked on the constraint + half the buffer

If that falls within your standard lead time, promise it with confidence. If it doesn’t, don’t guess: quote the safe date, or add capacity at the constraint first.

When the planned load is low, you have room to take rush orders, often at a premium. When it nears the standard lead time, it is an early warning: act on capacity before the dates start slipping.

The TOC way of production, in one picture

Flow is the first objective of operations.
Over-production is prevented by a practical mechanism: release is restricted, so material can’t arrive ahead of time.
Local efficiencies are abolished, and synchronisation comes from releasing in step with the constraint.
A focusing process balances the flow: buffer management and load control, with the reasons for blacks and reds guiding improvement.

Signs that it’s working: on-time delivery against the original promise above 95%, no month-end dispatch rush, a visible and stable constraint, and non-constraints that sometimes stand idle without anyone panicking.

Check your understanding

Answer all five. Get four right to complete the lesson.

1. The fatal assumption in MRP is…

That customers order in advance
Fixed lead times that ignore capacity and queues
That inventory is free
That there is one constraint

Lead time depends on the work ahead in the queue, which MRP doesn’t see.

2. Why does frequent rescheduling hurt on-time delivery?

It costs software licences
It makes the plan too simple
It needs more planners
Each reschedule moves promised dates and spreads local disruptions across the plant

TOC buffers the plan instead of rewriting it.

3. A safe due date is roughly…

Today + work already booked on the constraint + half the buffer
Today + the customer’s wish
The end of the month
Today + the sum of all operation times

It reflects the actual load, protected by half the buffer.

4. With TOC, accurate capacity data is needed mainly…

At every machine
At the suppliers
At the constraint
Nowhere

Once the release is choked, one constraint governs the plan.

5. Which is a sign the TOC way of production is working?

Every machine is busy 95% of the time
On-time delivery against the original promise above 95%, with no month-end rush
The plan is rescheduled daily
Inventory is at its highest

Reliable promises and steady output are the markers.