A New Planner's First Week: From First Sign-In to a Schedule the Shop Trusts

A new planner can go from first sign-in to a committed schedule in five days, and the order matters: read the plant, read one job, run a small selection, log actuals, then commit. This walkthrough follows a planner's first week at Halden Precision, a 34 job shop with 11 work centers and two shifts, learning EDGEBIC by User Solutions from scratch. It is one of the worked examples we publish so the sequence is visible rather than described.

The mistake almost every new planner makes is running the scheduler on day one and then spending four days arguing with dates they cannot explain. Reading first is faster.

Day 1: Read the Plant, Not the Plan

Sign in and stay out of the schedule entirely for the first morning.

The main window is one row of tabs, and they sort into three jobs. Customers, Daily Hours, Product, Workcenter and BOR describe your plant. The Schedule tab is where jobs get planned and read. Dashboard, Reports and Settings are analysis and administration.

Open them in the order the plant depends on them.

Daily Hours, Shift tab. Halden runs Day Shift 06:00 to 14:30 and Afternoon Shift 14:30 to 23:00, Monday to Friday. The grid shows both, with weekly totals. Read every weekday row. The scheduler believes this calendar literally: if a shift says Saturday runs, jobs can be placed on Saturday, and if it says the day ends at 14:30, nothing is placed after 14:30.

Daily Hours, Holiday tab. Four plant holidays are listed, two of them recurring. One is missing: the plant closes the Friday after Thanksgiving and there is no record of it. Note it.

Workcenter tab. Eleven rows, with columns for instances, shift calendar (Global or Custom), and One Per Day. Two findings in the first hour:

• The bandsaw is listed with 1 instance. There are two saws on the floor. That machine has been planned at half its real capacity for months.
• The deburr bench shows a Custom shift calendar, which means plant-wide shift edits never reach it. Nobody remembers why.

This is the highest value hour of the week. Every strange date you will see on Wednesday is created here.

Day 1, Afternoon: One Routing You Know

Pick the product you understand best and open its routing on the BOR tab.

Halden's planner picks a hydraulic manifold: four steps, saw, mill, deburr, inspect. Read the numbers on each step and compare them against what the shop actually does.

One number is wrong and it is a big one. A 0.8 hour setup understatement on a machine that sees six setups a week is nearly five hours of invisible load. Note it; do not change it yet. Day 1 is a reading exercise, and you want the corrections reviewed by whoever owns the routing.

Day 2: Read One Job End to End

Open Schedule then Job View and pick a single live job.

The grid gives one row per operation, with Sched Start, Sched End, Sched Hours, Actual Start, Actual End, Rem. Hours and Percent Complete. Beside it the Gantt draws the same plan as bars.

Walk the job by hand. Step 10 ends Tuesday 11:20; step 20 starts Tuesday 11:20. Step 30 does not start until Wednesday 06:00, and the reason is that step 20 ends at 22:40 and the deburr bench has no afternoon shift. That is not a bug, that is the Custom calendar you noticed yesterday.

Then read the three job level numbers on the same tab: Item Start is when the last work center operation ends, Job End is Item Start plus the product's lead time in calendar days, and Due Date is the promise. Days Late compares Job End to Due Date, and it turns red before the due date passes, which is a warning rather than a history entry.

By the end of day 2 you can explain every date on one job out loud. That is the whole skill.

Day 3: Your First Run, Five Jobs Only

Open Drive Schedule. Tick five jobs you understand. Click Schedule + Re-Schedule.

The confirmation dialog tells you exactly what is about to happen, in the form "you are about to schedule 5 jobs: 2 new, 3 existing". Say yes. A progress overlay runs, and cancel is always safe because nothing partial is kept.

The critical property to internalize: a run touches only the jobs you ticked. The other 29 jobs keep their plans and their reserved capacity, and your five jobs are placed around them. That is why a small first run is safe.

Now verify by hand. Four of the five land where you expect. The fifth pushes out three days.

Work the standard checklist in order. Cause 1: someone else got the capacity first. Open the work center utilization report, drill into Mill-2, and there it is: a priority 1 job holds the mill Monday through Thursday. Your job is priority 5. The date is correct; the assumption was wrong.

That single trace is the day's real lesson. Reading the mechanics is in how to run the work center utilization report and the sequencing rules are in how to set a job priority.

Day 4: Actuals, and Why They Come First

Spend the morning at the kiosk with two operators, watching them punch in and out against operations. Then spend the afternoon in the Log Actuals grid seeing what those punches became.

The grid holds one row per calendar day of an operation, with Sched Hrs beside Actual Hrs. Days the system backfilled from the plan carry an auto-filled badge, and typing a real value over one clears the badge. Yesterday's saw operation shows 5.0 planned and 6.5 logged.

Now reschedule that one job and watch what happens. The saw step is completed, so its dates and hours are written back exactly as logged and never recomputed. The mill step is not started, so it re-plans forward from the saw's real finish rather than its planned finish, and it moves 1.5 working hours later. The job end moves with it.

This is the rule that makes the rest of the system worth trusting: completed work is never moved by a reschedule. The full guarantee is in why completed work is never moved on reschedule, and the daily habit is in how to enter actual hours for yesterday.

Day 5: Reports, Then the First Real Commit

Open the Reports hub. Every report queries fresh when you open it and nothing runs in the background. Every report also carries a Column Details button that explains its own columns with a formula, a unit, and example values, which is faster than asking a colleague and produces the same answer every time.

Four reports carry a new planner's first month:

Then commit. Fix the two master data errors you found on Monday, which is the second saw instance and the missing Thanksgiving Friday holiday. Leave the mill setup time alone until the routing owner confirms it. Run the plant. Read Late Jobs once more.

Halden's numbers on Friday afternoon: 34 jobs, six late, all six with a named cause. Before the two master data fixes, the saw looked 98 percent loaded and was the apparent bottleneck. After them it reads 49 percent, and the real constraint turns out to be Mill-2 at 91 percent. One number on one screen moved the plant's bottleneck.

The Week, Compressed

1. Read the calendar before anything else. Every date in the system is computed from shift hours and holidays.
2. Count machines against instance counts. It is the single most common master data error, and it moves the apparent bottleneck.
3. Explain one job's dates out loud before you run the scheduler. If you cannot, you are not ready to judge a hundred jobs.
4. Run five jobs, not the plant. A run touches only what you tick.
5. Log before you reschedule. The engine plans forward from your last logged position, so stale actuals produce a confident, wrong plan.

For the rhythm that follows this week, read a day in the life of a planner. For the underlying mechanics of why dates land where they do, finite versus infinite capacity scheduling is the concept, and the scheduling engine guide is the full pipeline.