Outgrowing the machine-shop Excel schedule
Most shops plan in a spreadsheet, and the research says most spreadsheets contain errors nobody has found. This is about where the planning sheet stops working, what the evidence actually supports, and how to tell which side of the line you are on.
Search for machine shop planning software and the results are not what a vendor would hope for. Alongside the tool searches sit “machine shop capacity planning excel”, “machine shop scheduling excel template”, “production planning excel sheet download”. The thing people want to replace is usually not other software. It is a file — and a good share of those searches are after a better version of the file rather than an escape from it.
That is worth taking seriously rather than sneering at. This is an attempt to be precise about where the spreadsheet stops: what the evidence supports, what it does not, and which failures are properties of the tool rather than of the planner.
The starting position, in numbers
The spreadsheet is not a fringe choice. It is the median one.
- 54% of plants globally still managed manufacturing operations with a mix of pen, paper and spreadsheets in 2024 (IoT Analytics, 2025).
- Among small and mid-sized manufacturers already shopping for software — a group primed to change — 23% plan in spreadsheets and another 25% by manual methods (Capterra, 2026).
So if your plan is an .xlsx with a tab per month and colour codes documented nowhere, that is not backwardness. The question is only whether it still fits your shop.
The part nobody checks
Here is the finding that should give any spreadsheet-based planner pause, and it is not about manufacturing at all — it is about spreadsheets.
When operational spreadsheets are inspected properly — several days per file, usually by commercial auditing firms, in one case by a tax authority doing cell-by-cell inspection — errors turn up in 94% of them, across 85 spreadsheets studied (Panko, 2015). Earlier field audits using weaker methods found errors in 24% of 367 spreadsheets; the more rigorous the audit, the higher the number, which tells you what the lower figures were really measuring (Panko, 2000).
The second finding matters more than the first. In an experiment where developers were asked how likely it was that they had made an error, the median estimate was 10% — while 86% of them had in fact made one (Panko, 2015, reporting Panko & Featherman 1999). Those subjects were students, so treat the exact figures with care; the direction is consistent with decades of work on overconfidence.
Put those together and the practical conclusion is uncomfortable but simple: the sheet is probably wrong somewhere, and you almost certainly do not know where. For a planning sheet, an undetected error is not a rounding problem. It is a job that quietly is not where you think it is.
Where it breaks
Four failures follow from what a spreadsheet is, rather than from how carefully it is kept.
Conflicts stay invisible. Two jobs on the same machine in the same week are two rows, or two cells in adjacent columns, and nothing about them is red. A spreadsheet stores what you type; that two jobs need one physical thing at one time is a fact about the shop, not about the cells. Unless somebody writes the formula that says so — at which point they have started building scheduling software in Excel — the overlap surfaces physically, at the most expensive possible moment.
This is where the error research bites hardest. The check that would catch the collision is exactly the kind of formula that goes wrong unnoticed.
One person owns the file. The plan lives on a laptop; everyone else works from a printout or a copy that was current on Monday. Structure, colour codes and exceptions live in one head. That makes holiday a risk and handover a project.
Every change is a rebuild. A rush job displaces work, and what it displaced displaces more. The rebuild is mechanical and manual, and lands on whoever has least time for it.
Utilization is guesswork. A sheet records what was planned but does not accumulate, so “is the mill the bottleneck?” gets answered from memory. Underused capacity and chronic overload both cost money, and neither is visible until a date slips.
And a fifth that is less a break than a wall: a second site means a second file, at which point no sheet can answer the only question that matters — where should this job go?
What a replacement has to keep
This is where planning software usually loses, and why shops try one and go back to the file.
The spreadsheet got things right that are easy to lose. It kept the plan at a granularity a person can hold in their head — days, not minutes. It never overrode the planner. And it stayed cheap to change, which matters because no vendor’s data model survives first contact with a real shop floor.
A tool that optimises the plan into a shape the planner did not ask for and cannot explain to the floor has to be much better than a spreadsheet to survive, and usually is not.
What it has to add
Given that, the case for replacing the file rests on a short list.
Conflicts that surface when they are created — not on the floor, and not dependent on a hand-written formula being right. This is most of the value, because it converts a discovery problem into a decision.
One current plan. Reading it should not risk changing it, and the version somebody looked at ten minutes ago should be the version that exists.
Moves that carry their consequences. When a job moves, what it depends on moves with it, and anything that no longer fits says so.
Capacity that accumulates, so load is a number rather than an opinion.
For why the obvious answer — buy an ERP — does not apply at this size, the numbers are in Every factory has limited space: median enterprise software project $450,000 and 15.5 months, and ERP adoption at 41% of small EU enterprises against 89% of large ones.
One modelling point worth getting right
Something that trips up spreadsheets and software equally: what exactly are you scheduling?
The instinct is to keep a list of machines and, separately, a plan of floor space. Those are the same thing seen twice. A machine stands somewhere, that place has a footprint, and the job needs both at once. Two lists disagree, and now you are maintaining a reconciliation nobody asked for — another undetected-error surface, of exactly the kind the audit studies keep finding.
The version that stays consistent is to treat the workstation as the unit you plan. The mill, the press, the deburr bench, the assembly bay: each a place where work happens, with properties describing what stands there. Scheduling a job “on the mill” and booking the workstation the mill stands at are then one operation on one record. Portable things — fixtures, gauges — are genuinely separate, because they travel between workstations. Fixed machines do not.
How to tell which side you are on
The spreadsheet is still right for a shop where one person can see the whole plan, changes are rare enough to absorb by hand, and there is one site.
The signals that you have passed that point are specific: someone walks to the floor to find out what is actually running; the plan has a single owner whose holiday is a risk; a rush job costs an afternoon of rearranging rather than a decision; nobody can say what last month’s utilization was; or you have opened a second site and started a second file.
If none of those are true, stay in the spreadsheet — and if you want a better one, ours is free. It flags overbooked days, which is roughly as far as a spreadsheet honestly goes, and it states its own limits on the front sheet.
If several are true, the file is costing more than it saves, and the answer is not a bigger spreadsheet. It is scheduling that makes conflicts visible without taking the plan away from the person who owns it.
A note on the search data
The opening observation — that machine-shop planning searches skew toward spreadsheet templates — comes from our own keyword monitoring in August 2026, not from a published study. It is a small, short-window sample and should be read as an observation that matches the survey figures above, not as independent evidence.
Sources
- IoT Analytics, “MES Market 2025–2031: 300+ vendors replace pen & paper and spreadsheets” — research summary of the MES Market Report 2025–2031, 15 December 2025. 54% of plants globally in 2024 used a mix of pen & paper and spreadsheets to manage manufacturing operations.
- Capterra, “MRP Software vs. Spreadsheets” — analysis of software-buyer conversations with small and mid-sized manufacturers, 1 Jan 2025 – 1 Jan 2026 (n = 1,848). Capterra states the findings “represent buyers who contacted Capterra and may not be indicative of the market as a whole” — a sample already leaning toward software adoption, which if anything understates spreadsheet reliance.
- Panko, R. R., “What We Don’t Know About Spreadsheet Errors Today: The Facts, Why We Don’t Believe Them, and What We Need to Do” — Proceedings of the EuSpRIG 2015 Conference, Spreadsheet Risk Management, ISBN 978-1-905404-52-0. Table 2: 85 operational spreadsheets subjected to intensive inspection, errors found in 94%. The overconfidence experiment (median self-estimated error likelihood 10%, actual 86%) is Panko & Featherman 1999, reported in this paper; its subjects were students.
- Panko, R. R., “Spreadsheet Errors: What We Know. What We Think We Can Do.” — Proceedings of the EuSpRIG Spreadsheet Risk Symposium, Greenwich, July 2000. Table 1: seven field audits, 367 spreadsheets, errors in 24% overall; audits from 1997 onward, using stronger methods, found errors in at least 86% (91% across the 54 spreadsheets in that subset). Note the age of this paper — it is included because it shows how the figure moves with audit rigour.