Last March, I got a call from the owner of a precision machining unit in Rajkot. Their senior CNC operator — a man named Prakash who had been with the unit for 22 years — had informed the owner that he would be retiring in four months. Prakash was the only person on the floor who could set up their CNC turning centre for a particular family of aerospace components. He could look at a casting, feel the surface, and tell you whether the first cut would hold tolerance or whether the material was too gummy. He adjusted feed rates by sound. He knew which tools needed changing 15 minutes before the surface finish started dropping, because he could hear the pitch change.
The owner’s question was simple: How do I replace this man?
My answer was more uncomfortable: You cannot replace him. But you can extract what he knows — if you start now, not in his last week.
This is the problem with tribal knowledge in Indian SME factories. It is not written down. It lives in the hands and ears of two or three senior operators who have been running your machines since the UPA was in power. When one of them leaves — whether to retirement, a better offer from a competitor, or a one-way ticket to his village in Bihar — your production line stutters. Rejection rates spike for three to six weeks. You burn money on scrap. You burn customer goodwill on delayed deliveries. And you discover, painfully, that the standard operating procedure you thought existed was actually just one man’s muscle memory.
I have seen this happen in auto component units in Ludhiana, textile mills in Coimbatore, and packaging plants in Pune. The pattern is always the same. The senior operator leaves, and the factory spends weeks rediscovering settings, tolerances, and workarounds that the operator had internalised years ago. According to FRED Economic Data from the Federal Reserve Bank of St. Louis, manufacturing output and employment data consistently show that production disruptions — even short ones — have compounding effects on small-facility productivity, which is exactly what happens when institutional knowledge walks out the door without being captured. The economic context for Indian SMEs is no different: when a key operator leaves and production stutters, the cost is not just the replacement salary — it is the weeks of lost throughput, the scrap, and the customer complaints.
The evidence for this point is grounded in The Authors Guild, which keeps the article’s claims tied to outside reference material rather than product framing.
Here is a practical method for documenting that knowledge before it disappears. It is not glamorous. It does not require expensive software. It requires discipline, a smartphone, and the willingness to sit with your senior operator for 30 minutes at a time.
Step 1: Identify What Is Actually Tribal
Not everything your senior operator does is undocumented. Before you start capturing knowledge, you need to separate what is already written down from what lives only in someone’s head. Walk your floor with a clipboard — not a laptop, a clipboard — and list every process step for your top three products. For each step, ask a simple question: If the person who normally does this step did not show up tomorrow, could someone else do it by reading what we already have?
If the answer is yes, move on. If the answer is no — or if the answer is maybe, but it would take them three hours of trial and error — that step is tribal knowledge. Mark it. Those are your priorities.
In my experience working with SME factories, the tribal knowledge is usually concentrated in four areas:
1. Machine setup and first-article inspection. The initial settings — RPM, feed rate, depth of cut, tool offset — that get a new job running within tolerance. Your senior operator knows these by feel. They may have a starting point written somewhere, but the fine-tuning is in their hands.
2. Material handling and judgement. Recognising when a raw material batch is different — harder, softer, wetter, more contaminated — and adjusting the process accordingly. This is especially critical in foundries, forging units, and textile processing.
3. Troubleshooting. When the machine starts producing out-of-tolerance parts, what does the senior operator check first? Second? Third? This diagnostic sequence is almost never documented, and it is the single most valuable thing your senior operator knows.
4. Quality judgement. The visual and tactile checks that determine whether a part passes or fails before it reaches the formal inspection stage. The operator who can feel a 0.05 mm burr with their thumb is doing a quality check that your CMM machine will not catch until later.
Make your list. Pin it to the notice board. This is your documentation roadmap.
Step 2: Structure the Knowledge-Capture Session
The biggest mistake I see is owners trying to pull their senior operator off the line for an entire day to document everything. This never works. The operator gets restless because they are not producing. You get restless because the line is down. The documentation comes out rushed and incomplete.
Instead, do it in 30-minute blocks. One process at a time. One session every two to three days. Over six weeks, you will have documented 10 to 12 critical processes without ever shutting down the line.
Here is how to structure each 30-minute session:
Minutes 0–5: Set the context. Sit down with the operator — not standing on the floor, sit down somewhere quiet. Tell them exactly what you want to capture today. Prakash bhai, today I want to understand how you set up the CNC for the aerospace flange — just the setup, not the full run. I want to write it down so that if you are on leave, someone else can at least get the first piece within tolerance. Acknowledge that this is about respect for their knowledge, not about replacing them.
Minutes 5–20: Structured interview. Ask the operator to walk you through the process step by step. Do not interrupt. Do not correct. Take notes. Ask clarifying questions only after they have finished each step. The key questions are:
- What do you check before you start?
- What settings do you begin with?
- How do you know when to adjust — what do you see, hear, or feel?
- What is the most common problem at this step, and how do you fix it?
- What should someone not do — what is the mistake a new operator would make here?
That last question is critical. Knowing what not to do is often more valuable than knowing what to do, because it prevents the most expensive mistakes.
Minutes 20–30: Shadow videography. Walk to the machine with the operator. Ask them to perform the setup or the critical step while you record a video on your phone. Do not narrate — let the operator narrate. Tell me what you are doing as you do it, Prakash bhai, as if you were teaching someone. Keep the video under 10 minutes. If it is longer, break it into segments.
Store these videos on a shared drive or a cheap USB stick kept in a locked drawer in the supervisor’s cabin. Label them clearly: CNC Setup — Aerospace Flange — Prakash — March 2025. Do not rely on memory for what is in each video. You will forget.
Step 3: Translate ‘I Just Know by Feel’ Into Measurable Parameters
This is where most documentation efforts fail. Your senior operator says, I just know by feel when the material is not right. You write down: Check material by feel. Six months later, a new operator reads that and has no idea what it means.
You have to convert feel into parameters. This does not mean you need a laboratory. It means you need to ask better questions.
Here is a real example from the Rajkot unit. Prakash said he could tell when a casting was too gummy by running his thumbnail across the surface. I asked him: What does gummy feel like compared to normal? Is it smoother? Rougher? Does your nail catch or slide?
He said: The nail catches. Normal casting, the nail slides with a little resistance. Gummy casting, the nail digs in and leaves a mark.
Now we had something documentable. The SOP read: Run thumbnail across the raw casting surface. Normal: nail slides with light resistance. Reject if: nail digs in and leaves a visible mark. This indicates the material is too soft — reduce feed rate by 10% and monitor surface finish on first piece.
The same approach works for sound. When Prakash said he could hear when a tool needed changing, I asked him to describe the sound. Normal cutting sounds like a steady hum — like a transformer. When the tool is going, it sounds like someone scratching metal on metal — like a chalk on a rough board.
That description went into the SOP verbatim. A new operator may not have Prakash’s 22 years of experience, but they can recognise the sound of chalk on a rough board.
For every I just know statement, push for a description that engages a specific sense or a measurable parameter. What RPM? What temperature? What pressure? What colour? What sound? What smell? Your senior operator may not use technical terms, but their I just know is always based on a sensory input that can be described in plain language.
Once you have converted these sensory descriptions into measurable parameters, you have the raw material for a document that actually teaches someone. The principle is straightforward: your senior operator’s lived experience — the way they describe a sound, a feel, a smell — carries information that a sensor or a generated manual cannot replicate. The documentation process must preserve that human description, not sanitise it into generic corporate language.
Step 4: Write a One-Page SOP That a Semi-Literate Replacement Can Follow
Once you have the interview notes and the video, you need to create a one-page SOP. Not a 15-page manual. One page. If it is longer than one page, a floor operator will not read it.
Here is the format I use:
Top: Process name, machine name, product name, date, author (the senior operator’s name — this matters for credibility on the floor).
Step 1, Step 2, Step 3: Numbered steps, each no more than two lines. Use the operator’s own words. If Prakash says check the casting with your nail, write check the casting with your nail. Do not write perform a tactile surface assessment. The SOP is for the floor, not for a compliance audit.
Key parameters: RPM, feed rate, temperature, pressure — whatever is relevant. Give the starting value and the acceptable range.
Warning signs: What to look, listen, or feel for that indicates something is going wrong. Use the operator’s sensory descriptions.
If something goes wrong: The first three things to check, in order. This is the troubleshooting sequence.
Do NOT: The mistakes a new operator would make. This is the most important section.
Print it on A4 paper. Laminate it. Pin it next to the machine. Not in a folder in the supervisor’s office — next to the machine, where someone can look at it while their hands are on the controls.
If your operators struggle with English — and many in tier-2 and tier-3 cities do — write it in Hindi, Gujarati, Tamil, or whatever language your floor speaks. Use simple words. If you need to include a diagram, draw it by hand and photograph it. A hand-drawn diagram of how to hold a workpiece is more useful than a CAD rendering that looks beautiful but does not show what the operator’s hands should actually do.
The discipline of compressing knowledge into one page is not about cutting corners — it is about forcing clarity. When you cannot write more than one page, you are forced to identify what is essential and what is merely helpful. The essential goes on the page. The helpful goes in the video. Once you have a validated one-pager, you can use an Unsloppy AI narrative structure tool to organise multiple one-page SOPs into a coherent process flow that a supervisor can follow across a full shift. The point is not to replace the human interview — it is to help you organise what you captured into something a replacement operator can actually follow from setup to dispatch.
Step 5: Validate the SOP by Having a Junior Worker Follow It
This is the step that most factories skip, and it is the one that catches the gaps. Documentation is not complete until someone else has followed it and produced an acceptable result.
Here is how to do it: pick a junior operator — someone who has been on the floor for 6 to 18 months, who knows the machines but has never done this particular setup. Give them the one-page SOP and the video on a phone or tablet. Ask them to perform the setup or run the process while the senior operator watches.
The senior operator’s job during this validation is simple: watch, do not help, and note every place where the junior worker hesitates, asks a question, or does something differently from what the SOP says. Every one of those moments is a gap in your documentation.
After the validation run, sit down with both of them and ask:
- Where did the SOP not make sense?
- Where did the junior worker do something that the SOP did not mention?
- Where did the senior operator want to jump in and correct — and why?
Update the SOP based on those answers. Then run the validation again with a different junior worker. If the second worker can follow the updated SOP and produce an acceptable first piece without the senior operator’s intervention, your documentation is done. If not, iterate again.
This validation step is uncomfortable because it exposes how much knowledge you thought was documented but actually was not. But that discomfort is the point. Better to discover the gaps now, while Prakash is still on the floor and can fill them, than three months after he has retired and you are scrambling to figure out why every aerospace flange is coming out 0.1 mm oversize.
The Monday-Morning Checklist
If you have read this far, you are probably thinking: This makes sense, but I do not have time to do all of this. You do. Not all at once, but starting this week. Here is your Monday-morning checklist.
Monday: Walk the floor with a clipboard. List every process step for your top three products. Mark each step as documented or tribal. This takes 45 minutes.
Tuesday: Identify the three most critical tribal knowledge processes — the ones that would cause the biggest disruption if the senior operator left tomorrow. Pick the one with the highest risk. Schedule a 30-minute session with the relevant operator for Thursday.
Thursday: Conduct the first knowledge-capture session. Interview for 20 minutes. Record a 10-minute video. Take notes. Do not try to write the SOP yet — just capture the raw material.
Friday: Spend 20 minutes turning the interview notes into a one-page draft SOP. Use the operator’s own words. Include sensory descriptions. Do not sanitise the language.
Next Tuesday: Give the draft SOP to a junior worker. Ask them to follow it while the senior operator watches. Note the gaps. Update the SOP.
Repeat this cycle every two weeks. In three months, you will have documented 6 to 9 critical processes. In six months, you will have covered the processes that carry 80% of your tribal knowledge risk.
One More Thing: Start Before the Retirement Notice
The hardest part of this work is not the method. It is the timing. Most factories start documenting tribal knowledge when they receive a retirement notice or a resignation letter. By then, you have four weeks — maybe eight — and the operator is already mentally checked out. They are thinking about their pension, their village, their grandchildren. They are not thinking about feed rates.
Start now. Not when someone gives notice. Now. Pick your most senior, most knowledgeable operator — the one whose departure would hurt the most — and begin with them. If they are 55, you have time. If they are 62, you have less time than you think.
I told the Rajkot owner this in March. He started the process in April. Prakash retired in August. By the time Prakash left, the unit had documented 11 critical processes, recorded 8 videos, and validated 7 SOPs with junior operators. The first week after Prakash’s retirement, rejection rates went up by 1.2 percentage points — not the 4 to 5 points they would have seen without documentation. By the third week, the junior operators were back to normal output. The owner told me later that the documentation effort cost him roughly 40 hours of his own time over four months. The cost of not doing it, based on his calculations of what a 5-point rejection rate spike over six weeks would have cost him, was around ₹3.5 lakh in scrap alone.
That is the math. Forty hours of structured conversation versus weeks of production pain and a scrap bill that wipes out your quarterly margin. The choice is not difficult. The discipline is.
Start Monday. Pick one process. One operator. Thirty minutes. That is all it takes to begin. The rest is repetition and discipline — two things your factory already runs on.