A Customer Asks to Pull In a Delivery: Answering It in 20 Minutes

A customer pull-in request is answerable in about 20 minutes: change the due date, re-plan the job, read Days Late, and if it is not zero, work three levers in order until it is or until you have a defensible no. This walkthrough runs that loop in EDGEBIC by User Solutions on a real job with real numbers, including the part most articles skip: checking what the pull-in did to everybody else. It is one of the worked examples we publish so the mechanics are visible rather than asserted.

The value here is not that the software says yes. It is that it says yes or no in minutes, with the arithmetic attached, instead of a planner spending an afternoon in a spreadsheet and still guessing.

The Cast: One Job, One Phone Call

Ridgeline Metalworks has job J-4417 in the plan: 60 welded frames, currently scheduled to finish Thursday, October 16, against a due date of Friday, October 17. One day of float. A comfortable job.

Then the customer calls. Their install crew moved up, and they want the frames on Friday, October 10. That is five working days earlier.

The routing:

All work centers run one 8 hour day shift, Monday to Friday. Nothing on this job has started yet.

Note that the current end date is Thursday October 16 rather than Tuesday October 14: the frame carries a two day post-production tail in its product lead time, which EDGEBIC adds after the last operation when it computes the job end. That tail is real and it is part of the promise, which is why the pull-in has to beat it too.

Step 1: Change the Date, Do Not Change the Plan Yet

Open the Schedule area, go to the Drive Schedule tab, and find J-4417. Click into the Due Date cell and type the new date, Friday October 10. The Save Changes button appears on the filter bar. Click it.

Nothing on the Gantt has moved. That is deliberate: an inline edit changes the promise, not the plan, and the plan only changes when you run the scheduler. It means you can stage several date changes and evaluate them together rather than watching the schedule rewrite itself under the shop's feet. The field mechanics are in how to change a job's due date.

Step 2: Re-Plan the One Job and Read Days Late

Tick J-4417 only, and run the scheduler. Scoping the run to the one job leaves every other job's reservations in place, so the pull-in has to fit around real commitments rather than into a fantasy empty plant.

The result: Days Late reads 4.

The job still ends Thursday October 16, because nothing about the plan changed. The due date moved and the work did not. Four days is the honest gap between what the customer wants and what the plant currently does.

This is already worth the 20 minutes. You now know the size of the problem, and you know it before saying anything to the customer.

Step 3: Find Where the Four Days Actually Live

Open the job and look at the sequence. The saw and the mill both start on the first day they could. The weld step is where the time is: 22 hours on a single-instance cell over three working days, and it cannot start until the mill finishes Wednesday.

So the four days are not spread evenly. They are concentrated in one 22 hour operation on one machine, plus the two day product tail that cannot be compressed at all. That narrows the levers considerably.

The general diagnosis path is how to see why a job is late.

Step 4: Lever One, Priority

The cheapest lever is order of service. If J-4417 is currently priority 30 and two jobs ahead of it on MILL-2 and WELD-A sit at priority 20 with days of float against their own due dates, moving J-4417 to 15 lets it claim the earlier slots.

Change the Priority cell to 15, save, and re-plan the affected jobs.

Result: the mill step pulls in one day and the weld step starts Wednesday instead of Thursday. Days Late drops from 4 to 3. One day recovered for free.

Priority is genuinely free only when the jobs it displaces have float. Check them. If displacing them turns two other orders late, you have not gained a day, you have traded one customer's problem for two others'. Number in bands (10, 20, 30) so a rush insertion is one edit rather than a renumbering exercise, as covered in how to set a job priority.

Step 5: Lever Two, Capacity on the Constraining Machine

Three days remain and they sit on WELD-A. The weld cell has no shift on Saturday.

Add a capacity override of 8 hours on WELD-A for Saturday October 11 and Saturday October 4. An override is read ahead of the shift formula for that specific work center and date, and when it covers a day that has no configured shift the engine creates a synthetic override-day slot to hold the work. It is a one-time change for two dates. The recurring calendar is untouched, so nobody comes in on the following Saturday by accident.

Re-plan. The weld step now runs Wednesday, Thursday, plus the Saturday override, finishing Saturday afternoon instead of Monday. Days Late drops from 3 to 1.

The mechanics of the override, including why it works on a day with no shift, are in running a weekend on a capacity override.

Step 6: Lever Three, Route Around the Constraint

One day left. WELD-A has a sibling, WELD-B, which is bound into the same work center group. If the weld step is bound to the pool rather than to the single machine, the engine shops both members and picks by the group's strategy.

Bind step 30 to the pool and re-plan. Under earliest completion, the engine splits nothing but it does choose the member that finishes soonest, and WELD-B has a free Thursday that WELD-A does not.

Days Late reads 0. Job end lands Wednesday October 8, plus the two day tail, which is Friday October 10. The requested date, exactly.

Step 7: Check What It Cost Everyone Else

This is the step that separates a real answer from an optimistic one. Run the plant-wide plan and look at the grid, not just at J-4417.

One casualty. J-4425 lost its single day of float to the priority change. That is now a decision, not an accident: either J-4425's customer can absorb a day, or the pull-in gets declined, or J-4425 gets its own override.

Then check the load. For the pull-in week, the weld cells run at 88 percent utilization. That is tight but survivable: one bad morning does not automatically break the promise. Had it read 99 percent, the honest answer to the customer would be that the date is achievable only if nothing goes wrong, which is a different sentence from yes. Running that check is how to run the work center utilization report.

What This Walkthrough Proves

1. Days Late is the answer, not an opinion. It compares the planned job end against the due date, and it reads 0 or it does not.
2. The gap has a location. Four days spread across a routing is a different problem from four days sitting on one machine, and the plan tells you which one you have.
3. The levers have an order. Priority is free when float exists, capacity costs money, routing changes cost nothing but need a pool to exist.
4. A pull-in is a trade. Checking the rest of the grid afterward is what makes the yes defensible.
5. Twenty minutes beats an afternoon. The customer gets an answer while they are still on the phone, and it comes with the arithmetic behind it.

The mirror image of this journey, an order that arrives new and urgent rather than an existing one moving earlier, is adding a rush order to a full schedule. For promising a date on work that is not yet an order at all, quoting three routing options for a customer uses quote simulation instead.