Working the Optimizer Into Your Scheduling Routine

Run the optimizer after every meaningful scheduling run, pick the goal that matches the week rather than the habit, and read the move list before you Accept. That is the whole routine. A run costs the time budget you give it, typically thirty seconds, and it cannot change anything without a deliberate click, so the only real cost of running it is the half minute of reading.

EDGEBIC by User Solutions designed the optimizer as a second pass over a plan the scheduler has already built, not as a replacement for scheduling. This post covers where it fits in a normal week, which goal suits which situation, and how to keep the practice light enough that it survives a busy shift.

The daily pattern

The core loop is three steps.

Schedule. Run Generate Schedule as you always have. This produces the baseline: the plan built by taking jobs one at a time, most important first, placing every step at its earliest feasible slot.

Optimize. Open the Optimizer tab in Schedule Jobs, or click Optimize Schedule on the Drive Schedule toolbar, which switches you there. Pick a goal, pick a time budget of 10, 30, or 60 seconds, and run. The status shows that it is trying schedules, and Cancel stops it harmlessly.

Decide. Read the verdict sentence, the comparison table, and the move list. Accept or Discard. Only the operations in the move list ever move, and completed and in-progress work passes through unchanged either way.

Do it before you print or publish the day. Once the floor has the plan, the calculus changes and stability starts to matter more than the last few percent of improvement.

Choosing the goal for the situation

Four goals ship, and each expresses a different ranking of what matters.

On-time first is the default and the right choice most days. Fastest overall finish will accept a later finish on one job to shorten the whole plan, which is the correct trade when the plant as a whole is the constraint and the wrong one when a specific date is. Least setup time stays hidden until sequence-dependent setup data exists, since there is nothing for it to minimize before then.

Fewest changes deserves its own note. It improves the plan while disturbing the floor as little as possible, moving an operation only when the gain clearly justifies the disruption. That is the goal for Wednesday afternoon, not Monday morning. See the Fewest changes preset.

Every goal carries the same guarantee: never worse than your current plan on the terms you asked for. Trying one costs a run and a discard.

A week that works

Monday morning. Schedule, then optimize with On-time first, then publish. The plan is fresh, nothing is committed, and this is the run with the most room to improve.

Tuesday and Wednesday. Reschedule as actuals land, then optimize with Fewest changes. The floor is committed and stability is worth more than the last hour of makespan.

When a rush order lands. Reschedule so the order is in the plan, then optimize with Fewest changes and read the move list carefully. A mid-shift change is a real cost to a supervisor, and you want evidence it was worth it.

Before a capacity conversation. Run Fastest overall finish and note the makespan delta. If the plan cannot be compressed by sequencing, that is a strong signal the constraint is capacity or routing, not planning. See the makespan metric.

End of week. Run On-time first once more on the forward plan. If it reports no improvement, you go into the weekend knowing the plan survived a search.

Keeping the review to thirty seconds

The review is fast by design, and it stays fast if you read in this order.

Start with the verdict sentence, which states in plain words what improved and confirms that nothing already started was touched. Then read the top row or two of the comparison table, which are the measures your chosen goal ranks highest. Then scan the move list, which shows only changed operations with their old start, new start, and a short reason.

Three details make this quick. Change cells spell out the direction in words as well as an arrow, so no color decoding is needed. Unmoved operations are never listed, so a dozen rows on a two hundred operation plan is normal. And the safety strip under the result restates the three guarantees every time, so you never have to remember whether a run wrote anything.

Communicating the change

When you Accept something that moves a supervisor's day, tell them with the same words the system used. The Explain dialog gives a plain-language story in five sections covering the goal you asked for, how the search went, the outcome measure by measure, the most significant moves as sentences, and what was protected. A Copy button puts the whole story on the clipboard for an email.

That same text is what Accept writes into the audit record, so the conversation in the moment and the audit trail months later tell literally the same story. Nobody has to reconstruct why a plan changed.

Habits worth keeping

Run after every meaningful scheduling run. It is seconds and it is non-destructive.

Change the goal deliberately, per this week's priority, not out of habit.

Read the move list before accepting. You are the planner and the optimizer is your analyst. It reasons from master data, and master data never knows about the fixture out for repair.

Do not fight locked jobs. When the mathematical solver reports a job as reproduced exactly rather than rearranged, that is correctness first, and the Explain dialog names the reason per job. See how the optimizer locks features it does not model.

Fix master data before chasing optimizer results. An optimizer amplifies bad setup times and wrong calendars into confidently wrong schedules rather than correcting them.

The bottom line

The routine is schedule, optimize, decide, and it fits in the minute after a scheduling run. Use On-time first as your default, Fewest changes once the floor is committed, and Fastest overall finish when capacity rather than dates is the constraint. Read the verdict, the top measures, and the move list, then Accept or Discard, knowing that discarding costs nothing. To start, open the Optimizer tab in Schedule Jobs after your next schedule run. For the full picture, read the EDGEBIC optimizer guide and how to turn on the optimizer, and explore the platform at EDGEBIC.