When Should a Machine Shop Move Away From Paper?

The short answer: when finding a record costs more than creating it. In practice most shops reach that point somewhere between fifteen and thirty people, or when they take on a customer who audits them. The trigger is rarely the volume of records — it is the number of questions being asked about them.

Paper is not the problem

Worth being clear about this at the start, because the software industry has an interest in telling you otherwise.

Paper quality records are perfectly compliant. ISO 9001 does not require electronic documentation, and neither does AS9100. A well-run paper system with disciplined revision control and a tidy filing scheme will pass audit after audit. Some of the best-organized shops in the country run on paper and clipboards.

The problem with paper is not recording. It is retrieval. Writing a nonconformance takes four minutes. Answering “how many nonconformances did we raise on this part number in the last two years, what were the top three causes, and did the corrective actions hold?” takes a morning — and that is if the filing has been kept up.

Everything below follows from that distinction.

Five questions that tell you it's time

If any of these makes your stomach drop, you have probably reached the point.

1. “Show me the evidence for this clause.”

An auditor names a requirement and asks what records you hold against it. On paper, that means knowing where everything is and physically retrieving it while someone watches. If you prepare for audits by pulling files for two weeks beforehand, the system is costing you more than it should — you are simply paying in time rather than on an invoice.

2. “This gauge failed calibration. What did it measure?”

The one that keeps quality managers awake. A micrometer comes back out of tolerance, and every part accepted on that gauge since the last good calibration is potentially suspect.

On paper, answering that means cross-referencing calibration records against inspection records against job travelers, by hand, under time pressure, while product may be sitting at a customer. It is genuinely difficult and genuinely urgent, and it is the single strongest argument for connected records.

3. “Which revision was in effect when we made this lot?”

Document control is the most common source of audit findings, and the reason is almost always the same: a superseded revision stayed on the floor. Paper systems depend entirely on someone physically collecting the old copies. Every uncontrolled photocopy is a future finding waiting to be written.

4. “How are we trending?”

Clause 9.1.3 expects you to analyze and evaluate quality data. If defect data lives on paper, trending it means someone manually tallying records into a spreadsheet each month — which either does not happen, happens badly, or consumes a day that person does not have.

5. “Did that corrective action actually work?”

Effectiveness verification means going back weeks or months later and checking. Paper cannot remind you, so it relies on someone remembering. Usually nobody does, and the corrective action gets closed the day the action was completed — which is evidence it was done, not evidence it worked. Auditors know the difference.

The size thresholds, roughly

Not rules, but consistent enough to be useful as a sanity check.

Under 10 people, no certification requirement. Paper and spreadsheets are genuinely fine. Stay there. Buying software at this stage is spending money to feel organized.

10 to 25 people, pursuing or holding ISO 9001. Paper still works but is starting to cost real time. The tell is audit preparation — if it takes more than a couple of days, you are carrying a hidden expense.

25 to 50 people. Most shops in this band have moved or are actively frustrated. Multiple people are now creating records, so both consistency and retrieval degrade at the same time.

Over 50, or any aerospace work. Traceability requirements alone make paper expensive. AS9100 adds counterfeit part prevention, special process control with accreditation expiry tracking, first article inspection and configuration management — all cross-referencing problems, which is exactly what paper is worst at.

Two triggers that override size entirely

A customer audit. The first time a customer sends someone to audit your quality system, the cost of retrieval becomes immediate and visible. Shops that have been comfortably on paper for a decade frequently move within months of their first customer audit.

Aerospace or defense work. AS9100 traceability makes manual record-keeping painful at almost any headcount.

Before you move: three things worth trying first

Software is not the only answer, and some of the pain has cheaper fixes. If your problems are mild, try these before you spend anything.

• Consolidate to one location. A single controlled folder or filing cabinet, read-only where possible, with no local copies and nothing emailed around. Most revision-control problems come from parallel copies, not from paper itself.
• Build a review calendar. Calibration due dates, corrective action effectiveness checks, training expiry, internal audit schedule and supplier re-evaluation, in a shared calendar with named owners and reminders. Paper cannot prompt you; a calendar can.
• Index your records. A simple spreadsheet listing what exists and where it is turns a two-hour retrieval into a two-minute one. It is not elegant, but it addresses the actual problem.

If those three fixes make the pain manageable, you have bought yourself another year or two. If you try them and the retrieval cost is still unacceptable, that is a clear signal.

What moving actually involves

Less than people fear, and more than vendors admit.

You do not need to digitize your history

Start recording new work in the system from a chosen date and leave the old records where they are. Retention requirements are satisfied by the archive; nothing in the standard requires everything to live in one place. Shops that try to back-enter three years of records usually give up halfway and end up with two incomplete systems.

Expect the first month to be slower

People are learning, and entering things properly takes a few weeks to become habit. Plan for it rather than being surprised by it.

The thing that makes or breaks it

Whether the people creating records find the system faster than the clipboard. If they do not, they will quietly keep using the clipboard — and you will have paid for software and kept the paper problem.

This is worth testing before you buy. Have an actual inspector or machinist enter an actual record during the demo, not a salesperson. If it takes them longer than the paper form, that is your answer.

Run both for a while

For the first few weeks, keeping the paper form alongside the system costs almost nothing and gives you a fallback. Drop it once people stop reaching for it.

The economic question

The comparison people usually make is software cost against zero. That is the wrong comparison, because the current system is not free.

Add up the hours spent assembling evidence before each audit, chasing document revisions, cross-referencing calibration against inspection records, and rebuilding trend data by hand. Multiply by a loaded hourly rate. If that exceeds the annual cost of a system built for the job, paper is the more expensive option.

If it does not exceed it, stay where you are. That is a legitimate answer and it is the right one for a great many shops.

In summary

• Paper is compliant. The standard does not require software, and a well-run paper system passes audits.
• The trigger to move is retrieval cost, not record volume.
• The gauge-failure question is the sharpest test — if answering it would take you a day, connected records are worth real money.
• Under about ten people with no certification requirement, stay on paper and spreadsheets.
• A customer audit or aerospace work will bring the decision forward regardless of headcount.
• Try consolidating, indexing and a review calendar before you spend anything.
• When you do move, don't back-enter history, expect a slow first month, and test data entry with the people who'll actually do it.