On-time delivery:
See the delay while it is still avoidable
Which order will be late? Classification and survival models detect the looming delay early, while there is still time to prevent it, instead of just booking the penalty.
This module is part of a six-part model calculation about a fictional mid-sized manufacturing company (48 machines, 3-shift operation, €85M revenue). All euro amounts are derived from the documented assumptions of the sample plant. They are not results of real clients and not a performance promise. What values are achievable on your real data is established only in a pilot project.
01 The problem: You learn of the delay when it is too late
On-time delivery decides penalty and customer relationship
A late order costs twice: the agreed contract penalty and the customer's trust. Today the delay often shows only when the deadline can practically no longer be met. Then only expensive damage control remains.
The early indicators sit in the data: order lead time, material availability, current bottleneck load, backlog of upstream steps. A model estimates the probability of delay per order, weeks before the deadline.
Because one week of lead time is enough to reschedule capacity instead of paying the penalty.
02 The model: Delay risk per order
Classification and survival analysis on order progress, material and bottleneck load
Gefaehrdete Auftraege: 34 von 410 Mediane Vorwarnzeit vor Liefertermin: 9 Tage
The survival component delivers not just "on time/late" but how much buffer an order still has. This lets orders be prioritised: the one with the highest risk and the least remaining buffer comes first. That is data-driven sequencing instead of gut feeling.
03 Business impact: Avoid penalties, keep customers
The lever: avoided contract penalties and protected customer relationships (rush costs are counted in module 05)
Every assumption comes from the sample plant and is stored centrally. With your real figures only the input changes, not the method.
| Item | Value |
|---|---|
| Contract penalties / year | €340,000 |
| Avoidance via early warning (45%) | €153,000 |
| Protected customer retention (1 customer CM) | €95,000 |
| Result: savings / year | €248,000 |
Assumptions of a sample plant. In a real project your data replaces these values.
Contract penalties can be quantified, but the bigger value is often invisible: a customer who can rely on deadlines stays. Even one avoided supplier switch per year adds the contribution margin of a mid-sized customer to the bill.
04 Next steps on your shop floor
From the risk list to control
We connect order progress, material status and bottleneck load from ERP/MES. The fields already exist.
For one order segment the risk model is built. You see the early warnings validated against real past delays.
Daily risk list: "These orders are at risk. Recommendation: pull order X forward, reschedule capacity for order Y."