If It's Not Documented, It Didn't Happen

The FDA has a phrase every medical device manufacturer should take seriously: if it's not documented, it didn't happen. Your product may be brilliant. Your team may be experienced. Your factory may be spotless. None of that matters during an inspection if you can't prove — with timestamps, signatures, and revision history — exactly what happened.

This is where teams get stuck. An investigator asks for the device history record on lot #4721 and people spend two days pulling records from spreadsheets, paper travelers, email threads, and three disconnected systems. By the time the binder is assembled, the auditor has already written half their observations. Companies running an integrated system pull the same report in under a minute and hand over a clean, navigable record.

The gap between those two outcomes is rarely about effort. The first team works harder. They just work harder against a system that wasn't designed to produce evidence on demand.

There are three failure modes we see again and again, and they tend to compound.

First, manual data entry creates a lag between what happened on the floor and what's recorded. An operator finishes a step at 9:14 AM, but the paper traveler doesn't get keyed into the system until end of shift — assuming the form gets entered at all. That lag is invisible most of the time. During an audit, it becomes the question you can't answer: was the in-process check actually performed before the next operation began?

Second, inconsistencies between systems create gaps that raise red flags. The ERP says one thing about the BOM. The PLM says another. The MES has a third version. Someone built the unit using whichever printout was sitting on the bench. Reconstructing what was actually built — and proving it matched the approved revision — becomes archaeology.

Third, version drift in BOMs, work instructions, and SOPs makes it impossible to prove which revision was actually used at the moment of manufacture. "We know we use the latest one" is not a defense. The FDA wants the document under change control, the effective date tied to the work order, and the signature of the operator who acknowledged the revision.

Documentation debt accumulates silently. It looks fine at five employees and one product line. It still looks fine at twenty employees and three SKUs. Then a 483 lands, or you sign a hospital system that requires a quality audit before purchase orders move, or a competitor files a complaint that triggers an inspection.

Suddenly the same team that was "staying organized" is pulling all-nighters, hiring contract documentation specialists at $250/hour, and watching launches slip while engineers reconstruct device history records from memory and Slack messages.

We've seen companies spend six figures on a single remediation cycle that an integrated ERP would have prevented for a fraction of the cost. The avoided pain isn't visible on the P&L — but the inflicted pain absolutely is.

Documentation should happen as part of normal operations, not as a separate task your team has to remember. The operator scans the work order, the system shows them the controlled revision of the work instruction, and every action they take — material consumption, in-process check, equipment used, deviation logged — is captured automatically with a timestamp, a user ID, and a link back to the approved document.

Every transaction time-stamped. Every user identified. Every change traceable to an approved ECO. Every revision pulled from a single source of truth, with the older versions archived but never overwritten.

Done right, FDA submissions and inspections stop being stressful scrambles and become confident demonstrations of your quality system. The investigator asks a question. You click. The answer is on screen. The next question gets the same treatment. That's not luck. That's architecture.

If you're early — pre-commercialization, under 50 people, still on spreadsheets — the highest-leverage move is to choose a system that bakes in document control, electronic signatures, and lot/serial traceability before you scale, not after. Retrofitting compliance is roughly 5x more expensive than building on it.

If you're already scaling, the second-best time is now. Start with the workflows where a 483 would hurt most: device history records, design controls, CAPA. Move those into a controlled system first. The rest can follow.

Either way, the principle is the same. Stop asking your team to remember to document things. Build a process where doing the work is documenting the work. Then the FDA's golden rule stops being a threat — and starts being an asset.