Production Schedule

What this tool does

CM8-271 is a working production schedule: one row per job per work-centre operation, with planned dates and hours, a priority, a status, and — once the job finishes — the actual dates and hours it took. From those it measures the two things a schedule exists to protect: whether jobs finish when you said they would, and whether the hours you plan bear any resemblance to the hours you spend.

A schedule is a promise-keeping machine. Every planned finish date on it is a promise — to a customer, to the next operation, to the despatch bay. The sequence and the dates are only useful if the floor can actually meet them, and the adherence figures tell you, week after week, whether it can.

Everything runs inside this single file. There is no account, no upload and no network request of any kind, so your order book, customer names and delivery performance never leave the computer you are using.

Building a schedule the floor can meet

Most small-shop schedules fail the same way: they are wish lists. Every job is loaded at the date the customer asked for, nobody checks whether the hours fit, and the floor quietly decides the real sequence itself. The cure is finite thinking, even without finite-scheduling software:

• Load against capacity, by centre. The work-centre load chart sums the planned hours on each centre across the load window. You know roughly what each centre can do in that window — one machine, one shift, five days is about 40 hours. If the chart shows 57 hours of load on a 40-hour centre, the last 17 hours of promises are already broken; the only question is which ones.
• Schedule the bottleneck first. Every shop has one centre that is loaded harder than the rest — the one the load chart shows tallest, week after week. Sequence that centre first and fit everything else around it, because an hour lost there is an hour lost to the whole shop, while an hour lost on a lightly loaded centre usually costs nothing. Keep a little slack in front of the bottleneck so it never waits for work, and release jobs to the floor at the rate the bottleneck consumes them rather than the rate sales would like. That is the whole idea behind drum-buffer thinking, and it needs no software — just the discipline to let one centre set the pace.
• Plan in hours, not hope. A job with no planned hours cannot be loaded, and a centre whose load you cannot see cannot be protected. Estimate every job, even roughly — the variance figures will teach you to estimate better.

Priorities that mean something

The schedule has three priorities, and each one costs something:

• Normal — takes its place in the sequence. Most jobs, most of the time.
• Rush — a customer expedite. Rush is expensive: it breaks a setup, bumps other jobs, and usually makes at least one other promise late. Use it when a customer's line is down, not when a salesperson is nervous. If everything is rush, nothing is — the word stops carrying information and the floor goes back to choosing its own sequence.
• Firm — do not move. The material is committed, the promise is contractual, or the job lands in a window that will not come again (a planned shutdown, a fitting slot). Firm is sacred precisely because it is rare: it tells the planner which dates are load-bearing when everything else has to flex.

Statuses and the hold discipline

A job moves through planned (on the schedule, not yet issued), released (paperwork and material issued to the floor), running, and complete — or it is cancelled, or it stops at hold.

Hold is where jobs go to be forgotten, unless you make it painful. The tool will not save a hold without a reason, because a hold without a reason is a lost job: nobody knows what it is waiting for, so nobody is chasing it. "Material shortage — 3 mm sheet on back order" names the fix and the person who should be making a phone call. The on-hold tile shows the reason mix so the morning meeting can see at a glance whether the shop is waiting on suppliers, on quality decisions or on itself.

Measuring schedule adherence

A job is on time if it finished on or before its planned finish date — no grace days, no "nearly". The on-time tile and the monthly trend chart both use that definition, and the completed-jobs table shows the adherence of every recent job so a surprising rate can be traced to the jobs that caused it.

Adherence only measures history. The running late flag is the forward-looking half: any released or running job already past its planned finish is marked in the register and counted on the tiles, today, while there is still something to be done about it — a call to the customer, an overtime decision, a re-sequence. A late job you discover at completion is a statistic; a late job you catch mid-run is a choice.

Resist the temptation to re-date jobs to protect the rate. A schedule whose dates are quietly moved to match reality reports 100% adherence and tells you nothing. Move a date when the plan genuinely changes — and let the jobs that missed show as missed.

Hours variance — estimating feedback

Every completed job with planned and actual hours produces a variance: actual minus planned. One job's variance is noise — a tool broke, an operator was quick. The average across your recent completes is signal: if it sits persistently at +2 hours, every estimate you make is two hours optimistic, every quoted lead time is short, and every load chart understates the truth. Feed the number back into the next estimate and the schedule gets honest by itself. A variance near zero with scatter both ways is what good estimating looks like — not perfection on every job.

The daily ten-minute review

The schedule earns its keep in a short daily look, in this order:

1. Late right now. The tile and the register flag every released or running job past its planned finish. For each one: can it be recovered today, and who has to be told if it cannot? This list first, always — it is where promises are currently breaking.
2. Holds. Read the reasons. Anything held more than a couple of days needs a chase or a re-plan; a hold that nobody can explain gets released or cancelled.
3. Tomorrow's releases. Check the planned jobs due to start in the next day or two: material in? tooling ready? drawing issue resolved? Releasing a job the floor cannot actually start just converts a clean plan into a hold.

Ten minutes, every morning, beats an hour of replanning every Friday.

Schedule, capacity planner and WIP tracker

Three tools share this territory and do different jobs. The free production capacity planner answers the sizing question — how many hours each centre has against the demand on it, in aggregate, before any job has a date. The free WIP ageing tracker watches work already in progress and how long it has been sitting there. This schedule is the operational layer between them: the actual sequence of named jobs on named centres with dates, live status and the adherence record. Size capacity first, schedule jobs here, and let the WIP view catch what stalls.

FAQ

One row per job, or per operation? Per operation, where operations run on different centres — that is what lets the load chart see each centre separately. A three-operation job is three rows sharing a works-order number. If your jobs are single-operation, one row per job is fine.

What counts as on time — the day itself? Yes. Finished on the planned finish date is on time; the first day after is one day late. Adherence of zero or below is on time.

Should I put quoted delivery dates in as planned finishes? No — the planned finish is when the operation should finish, which needs to sit ahead of the delivery promise with time for whatever follows: later operations, inspection, packing, transit. If the operation date and the promise date are the same, there is no buffer, and every hiccup reaches the customer.

A job finished late but it was the customer's fault — does it count? Count it, and note why. The rate measures what the customer experienced. If customer-caused delays dominate, that is a finding worth taking to the sales meeting — and you can only take it there because you did not hide it.

Can a completed job stay marked rush? Yes — the priority is part of the history. Knowing which completes were expedites tells you what the expedites really cost in adherence and hours.

Saving your work

Jobs, settings and the report header are written to this browser's local storage as you type, and the toolbar shows the time of the last save. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the schedule for spreadsheet work and includes every filtered record, not only those drawn on screen. Reset asks twice, then erases everything this tool has stored. There is no undo.

Accuracy & disclaimer

This tool records intent and outcome and does the arithmetic between them faithfully. It cannot see a breakdown, an absence or a late delivery coming, it does not know how many hours each work centre actually has, and it cannot tell whether a planned date was realistic or hopeful. The adherence and variance figures are only as sound as the actual dates and hours entered against each job.

It is a scheduling record and a calculation aid, not a finite-capacity scheduling system, not a promise to any customer, and not a substitute for the daily conversation between the planner and the floor.