How to schedule a machine shop week in Excel without MRP

Why MRP is usually the wrong first step

MRP earns its keep when you are exploding multi-level bills of material, netting requirements across plants, and driving purchasing from a forecast. Most small machine and fab shops are not doing that. They are trying to answer three questions every Monday: what has to ship this week, which machine is the bottleneck, and which job is going to be late if nothing changes.

You do not need a system of record to answer those. You need one list that everybody trusts and a routine that gets it updated. A spreadsheet does that badly when it is three files that disagree, and well when it is one Excel Table with a fixed set of fields.

Step one: settle on one job list and one set of fields

Pick the smallest field set that answers a customer phone call. In practice that is job number, customer, part, quantity, promise date, current work center, remaining hours, and status. Anything else is nice to have and can be added as a column later.

Put it in an Excel Table, not a plain range. Tables keep formulas and formatting intact when rows are added, and they let you sort and filter without breaking the summaries that read from them. This is the single biggest reason shop spreadsheets rot: someone inserts a row and half the formulas stop pointing at the right cells.

• One row per open job, never one row per operation on a separate tab
• Promise date as a real date value, not text like "end of month"
• Remaining hours, not total hours — remaining is the number scheduling needs
• Status from a short fixed list, so it can be counted

Step two: sequence by promise date, then check the machines

Sort the list by promise date. That is your first-pass sequence, and for a lot of shops it is close to right. Then look at the hours the sequence implies for each work center over the next two weeks. If turning has sixty remaining hours against roughly seventy available and the laser has one hundred and ten against seventy, the sequence is fiction no matter how the dates look.

This is the whole point of a load view: it exposes the mismatch between the order you would like to run jobs in and the hours the shop actually has. You then do what shops have always done — move a job, add hours, subcontract an operation, or call the customer early. The difference is you are doing it on Tuesday instead of finding out on Friday.

Step three: flag risk with a window, not a gut feel

Pick a risk window in days — three is common — and let conditional formatting flag every job whose promise date falls inside that window while remaining hours are still significant. Past due gets its own colour. Nobody has to remember to check; the rows raise their own hands.

Keep the flag rules boring and visible. If the shop cannot explain why a row is orange, they will stop believing the colour.

Step four: publish one page

The schedule that gets followed is the one that fits on a page and gets printed or shared at the same time every morning. Filter to jobs at the work centers that are running today, sorted in sequence, with the flagged rows at the top. Print it, walk it out, and use it as the agenda for a five-minute meeting.

Do not distribute the whole workbook. One person owns the file and the floor gets the page. That single rule prevents the version sprawl that killed the last spreadsheet.

Step five: make the daily update short enough to survive

A scheduling routine dies when it takes forty minutes. Aim for five. The only daily entry should be current work center and status; remaining hours get touched when something changes materially. Weekly, review the flagged rows and the load view together and decide what moves.

If you want this already assembled — job list, work center load, at-risk flags, WIP summary, and a printable daily schedule wired together in one workbook — that is what the Production Control System is.