← Back to Field Notes

May 28, 2025 · The Bloomfield Team

How to Document the Process Only One Person Understands

A precision grinding shop in Pennsylvania lost their lead surface grinder in February 2024. He did not die or get injured. He retired. But the effect on the operation was almost as severe. He was the only person who could set up their Studer S33 for the aerospace bearing races that represented 22% of annual revenue. The shop spent four months and $180,000 in rejected parts, overtime, and outsourced work before a replacement was trained to an acceptable level. The knowledge to run that process had lived in one person's head for 18 years. Nobody had ever asked him to write it down.

Every manufacturing operation has at least one of these single-point-of-failure processes. The quoting methodology that only the senior estimator understands. The fixture setup that only the day-shift lead can do. The quality inspection sequence for the medical device customer that one person developed and nobody else has learned. Identifying and documenting these processes before the person leaves is one of the highest-value projects any manufacturer can undertake.

For a broader look at knowledge preservation, see our guide to manufacturing knowledge management.

Step 1: Identify the Single Points of Failure

Walk through your operation and ask one question at every station: if this person called in sick for two weeks, what would stop? The answers come quickly. The programmer who is the only one that post-processes five-axis code for the DMG. The maintenance technician who is the only one that can troubleshoot the PLC on the hydraulic press. The receiving inspector who is the only one certified to inspect incoming aerospace material.

List every process where one person leaving would cause a shutdown, a delay, or a quality risk. Rank them by business impact: revenue at risk, customer relationships affected, safety implications. Start documenting with the highest-impact process first.

Step 2: Record, Do Not Interview

The conventional approach is to sit down with the expert and ask them to explain their process. This produces a sanitized, simplified version that omits the critical details. People with deep expertise cannot fully articulate what they know because much of it is procedural memory. They do things automatically that took years to learn.

The better approach: watch them work and record what they do in real time. Video the setup. Photograph the fixture orientation. Note the sequence of steps, including the ones that seem trivial. When the grinder touches off and makes a micro-adjustment before the first pass, record what they looked at, what they felt, and what they heard that prompted the adjustment. Those micro-decisions are the expertise. The setup procedure is background.

Step 3: Capture the Why, Not Just the What

A procedure document that says "set the wheel speed to 6,500 RPM" is incomplete without understanding why that speed was chosen. The answer might be that the specific bearing race material work-hardens above 7,000 RPM, so 6,500 gives a margin. Or that the coolant flow at that station cannot keep up with the heat generated above 6,500 and the surface finish degrades. Those reasons are the knowledge. The setting is a fact.

For every step in the process, capture three things: what is done, why it is done that way, and what happens if it is done differently. Tribal knowledge is almost entirely "why" knowledge. The steps are visible. The reasoning behind them is not.

Step 4: Test with a Second Person

Documentation that the expert approves is not validated documentation. Documentation is validated when someone who has never performed the process can follow it and produce an acceptable result. Give the documented procedure to a competent operator who has not done this specific work before. Watch them follow it. Note where they hesitate, where they ask questions, and where they make mistakes. Each of those moments identifies a gap in the documentation that the expert filled unconsciously.

The first test will fail. That is normal. The gaps get documented, the expert fills them in, and the second test produces a better result. By the third iteration, the procedure is robust enough that a trained operator can follow it without the expert standing beside them.

Step 5: Build It Into the System

A documented procedure in a binder on a shelf is better than undocumented knowledge, but only marginally. The gap between documentation and usable knowledge is access. When the operator needs the information, they need it at the machine, in the moment, without searching through a filing cabinet or navigating a shared drive.

The documentation should be connected to the specific part number, machine, and operation in your system. When that job appears on the schedule and arrives at the work center, the setup notes, process warnings, and quality requirements appear with it. Converting machinist notebooks into searchable knowledge systems is how this knowledge survives beyond the person who created it.

The Timeline

Documenting one critical process takes 20 to 40 hours spread across two to four weeks. That includes observation sessions, documentation writing, expert review, and validation testing. For a shop with five single-point-of-failure processes, the total investment is 100 to 200 hours. Spread across six months, that is manageable for any operation.

The cost of not doing it is the $180,000 the grinding shop spent in four months of chaos after their expert left. Or the $1.2 million average cost of one retiring engineer when you account for lost knowledge, training time, reduced productivity, and quality issues during the transition.

The knowledge is in your building right now, distributed across people who are willing to share it if you build the process to capture it and the system to keep it alive.