Late vs. Overdue: Two Different Problems, Two Different Fixes

A job whose plan never reached the due date and a job whose plan was perfectly achievable but was not executed look identical on a simple late list, and they have no fix in common. Separating them is the difference between buying capacity you did not need and pressuring a floor that was never going to win. EDGEBIC by User Solutions splits them into distinct buckets, and the split changes who owns the row.

This post covers one specific return: the time and money you stop spending on the wrong fix. It sits under the EDGEBIC results guide and pairs with the true cost of a single late job, which prices what each row is worth.

One List, Two Unrelated Failures

Most plants have a late-job list. It is usually a single sorted list, and the weekly review works it top to bottom.

The problem is that at least two completely different failures land on that list, and the discussion has no way to tell them apart. So it drifts. Somebody argues the shop needs another machine. Somebody else argues the floor is not following the schedule. Both are sometimes right, about different rows, and the meeting ends without either being resolved because nobody separated the rows first.

Here are the two failures, stated plainly.

The plan never reached the date. Somebody promised a date, the routing and the available capacity do not add up to it, and the computed completion has always landed past the due date. Nothing on the floor went wrong. The arithmetic was wrong from the moment the promise was made.

The plan reached the date and was not followed. The computed completion did meet the due date. The date has now passed and the job is not done. The plan was achievable. Something stopped it.

No fix addresses both. Overtime does not repair an impossible promise, and a new machine does not explain why an achievable plan stalled for three days.

The Mechanism: Structural and Execution Lateness Get Different Buckets

On the CAPACITY tab's Overview, the Job Status donut buckets every active job into exactly one slice. Completed orders are excluded, so what you are looking at is the live population.

Two subtleties matter for how you read this.

First, whether a job has physically started is a badge on its row in the Active Jobs list, not a slice. A started job still lands in whichever bucket its dates dictate, because starting says nothing about finishing on time. Keeping production status and schedule health independently readable is deliberate.

Second, classification is judged against the delivery-ready date: the scheduled end plus the product's end-item lead time, which is the date the item is genuinely ready to deliver. That is why job rows show a ready date beside the due date, and why this view agrees with the Drive Schedule grid's own lateness column rather than contradicting it.

The Active Jobs list beneath the donut sorts worst-first, so it doubles as the attention queue.

The Causal Chain for Each Bucket

Trace both, because the intervention point is different.

Structural lateness. A date gets committed, often before anyone checks capacity. The routing books its hours against work centers that are already carrying work. Finite capacity places the operations honestly, into hours that exist, and the computed end lands past the promise. The job is flagged immediately. Every day it stays flagged, options narrow. The intervention belongs at the promise, and failing that, at the earliest possible replan.

Execution lateness. A date gets committed and the plan genuinely reaches it. Then the plan stops being followed: material did not arrive, a machine went down, an operator was pulled, a supervisor resequenced verbally, or the work happened and nobody logged it. The date passes. The job is flagged. The intervention belongs at whatever broke the link between plan and floor, and finding it means going to actuals, which is where a blocker hint that routes you to the right next report turns diagnosis into a lookup.

Which Fix Goes With Which Bucket

This is the whole practical payoff, so it is worth being explicit.

Notice the wrong-fix row, because that is where the money leaks. A shop that reads every late job as an execution failure ends up with a demoralized floor and unchanged delivery. A shop that reads every late job as a capacity shortfall ends up quoting capital projects to solve unlogged actuals. Both mistakes are entirely avoidable once the list is split, and neither is avoidable while it is one list.

Measuring the Split in Your Own Plant

You do not need a benchmark for this. You need a month of counting.

1. Take a snapshot each week of your active late population and record how many rows are structural and how many are execution.
2. Compute the ratio. Structural over total. This is the single most useful number the exercise produces.
3. Read the ratio as a diagnosis. Heavily structural means promise dates are being made without capacity behind them, which is a quoting and planning problem. Heavily execution means the planning is sound and something between plan and floor is not being captured or not being followed.
4. Price a sample of each. For a handful of rows in each bucket, note what you actually spent recovering them and what the cheapest fix would have been if the row had been diagnosed correctly on day one. The difference is the recoverable cost of misdiagnosis.
5. Watch the ratio move. Fixing quoting honesty should shrink the structural share. Fixing actuals discipline and floor handoffs should shrink the execution share. If neither moves, you are working on the wrong thing.

For documented outcomes rather than typical ones, the User Solutions and RMDB lineage includes GE Railcar moving from 30 percent to 90 percent on-time delivery, and Cummins deploying this scheduling approach across 33 locations. Those belong to their engagements and are quoted as heritage, not as forecasts.

Where This Distinction Does Not Help

Four honest limits, because a classification is only as good as its inputs.

A job with no due date cannot be classified. There is nothing to compare against, so it simply does not appear on the late list at all. If your late population looks suspiciously small, check for orders with empty due dates before congratulating anyone.

A wrong due date produces a confident wrong answer. If the date on the order is not the real customer promise, both buckets lie. A structural flag against a padded internal date wastes planning attention, and a clean bucket against an optimistic date hides a real miss. This is why honest promise dates are upstream of any lateness reporting.

A missing end-item lead time shifts everything. Because classification uses the scheduled end plus the product's lead time, a product whose lead time is unset will read better than reality and one whose lead time is overstated will read worse. Fix the product record before arguing about the bucket.

Execution lateness is invisible without actuals. If nobody logs hours, an achievable plan that quietly stalled looks exactly like one running fine until the date passes. The bucket will eventually be right and will have told you far too late. Kiosk actuals closing the planning loop is what makes the execution bucket timely rather than merely accurate.

Neither bucket tells you the cause. The split narrows the search from every possible explanation to roughly half of them. It hands you the right question, not the answer, and the answer still comes from the utilization report, the reschedule trail, or a conversation at the machine.

The takeaway

Two failures wearing the same label is why late-job reviews go in circles: the room argues about capacity and accountability at the same time, over rows that need opposite fixes. Splitting structural lateness from execution lateness turns one long circular meeting into two short specific ones, gives each row an owner, and produces a ratio that tells you where improvement money belongs. Count your own split for a month before you spend anything. To see the classification against your own orders, book a demo of EDGEBIC, and if you are on the older platform, moving from RMDB to EDGEBIC brings the dashboard with it. Read this alongside the true cost of a single late job and how EDGEBIC improves on-time delivery.