Last November, I stood on the shop floor of a packaging unit in Pune while the owner stared at a silent corrugator. The machine had tripped on a fault code nobody recognized. The only person who knew the reset sequence—a 58-year-old operator named Prakash—had retired the previous Friday. The owner had thrown him a small farewell party. Nobody had thought to ask Prakash to write down what he knew. The machine stayed down for three shifts. The cost of that downtime exceeded Prakash’s last six months of salary combined.

This is not a rare story. I have seen it in textile mills in Ludhiana, in auto-component shops in Coimbatore, in food-processing units in Nashik. Every manufacturing SME runs on knowledge that lives in the heads of a few people. The senior die-setter who knows exactly how many shims to add when the press starts drifting. The maintenance supervisor who can diagnose a bearing by touching the housing with the back of his hand. The founder who still remembers why a particular customer’s order always needs a different packing specification. When those people leave, the knowledge leaves with them. And the business pays for it in downtime, scrap, rework, and missed deliveries.

Most owners I meet treat this as an HR problem. They ask HR to “get an exit interview” or “collect the documents.” That approach fails because HR does not know what questions to ask, and the worker does not know what they know. Tribal knowledge is not a set of facts. It is a set of judgments, shortcuts, and pattern-recognition skills built over years. Capturing it is an operations problem. It requires the same discipline you apply to production planning, quality control, or preventive maintenance. This article is a field guide for doing exactly that—before your next Prakash walks out the door.

What Tribal Knowledge Actually Looks Like on Your Floor

Before you can capture something, you need to recognize it. Tribal knowledge in a manufacturing SME rarely looks like a missing manual. It looks like this:

• The extrusion operator who adjusts the temperature setpoint by 3 degrees on humid days, even though the SOP says 190°C year-round.
• The dispatch supervisor who knows that Customer A’s truck always arrives 90 minutes late, so he stages their load last—but the loading plan says first-in-first-out.
• The quality inspector who runs an extra check on batches from a specific raw-material lot because she remembers a rejection from three years ago.
• The founder who approves every purchase order above ₹10,000 because he knows which suppliers need a reminder call and which ones will deliver without follow-up.

None of this is written down. None of it is in your ERP. And none of it is transferable to the next person unless you make it transferable. The first step is to stop thinking of this as “experience” that cannot be documented and start thinking of it as operational risk that must be managed. Think of it like the approach outlined in the NIST Cybersecurity Framework: identify your critical assets, assess the risk of losing them, and build controls to protect them. In a factory, your critical assets include the undocumented knowledge that keeps production running. Losing Prakash is a risk event. You need a control for it.

Step 1: Identify Your Critical Knowledge Holders

Do not try to document everything from everyone. You will fail, and the attempt will annoy your workers. Instead, identify the five to eight people whose departure would cause a measurable disruption within one week. The test is simple: if this person did not show up tomorrow, what would break? Not what would be inconvenient—what would actually stop producing good output?

In a typical 50-worker factory, your list will probably include:

• One or two senior machine operators who run bottleneck equipment.
• The maintenance lead who knows the repair history of every major asset.
• The quality supervisor who understands customer-specific requirements that are not in the spec sheet.
• The production planner or scheduler who holds the real production sequence in their head, regardless of what the system says.
• Possibly the founder or plant manager, if they are still the only person who can make certain decisions.

Write these names on a whiteboard in your office. Next to each name, write the specific area where their knowledge is irreplaceable. “Ramesh—die-setting for Press 3 and 4.” “Fatima—packing specs for export orders.” “Venkatesh—boiler startup sequence after power outage.” This list is your knowledge-risk register. It tells you where you are exposed.

Step 2: Conduct Structured Walk-and-Talk Sessions

Do not hand these people a blank form and ask them to “write down what you do.” They will write what they think you want to read, or they will write nothing because they do not know where to start. Instead, you or a trusted supervisor need to walk the floor with them, observe them working, and ask specific questions.

The format I have seen work best is a 45-minute session focused on one machine, one process, or one customer type. Stand next to the person while they work. Ask:

• “Show me what you check before you start this machine.”
• “What sound or vibration tells you something is wrong?”
• “What do you do differently when the raw material comes from Supplier B instead of Supplier A?”
• “What is the most common problem you fix without anyone asking?”
• “If you were training someone to do this job, what three things would you make sure they never forget?”

Record the answers. A phone voice recorder is fine. A notebook is better because it forces you to summarize and clarify in real time. The goal is not a transcript. The goal is to extract the decision rules, the sensory cues, and the workarounds that the person applies without thinking.

One owner I worked with in a rubber-molding unit discovered during these sessions that his senior press operator had developed a method for detecting cavity-fill problems by watching the flash pattern during the first three cycles. The operator had never mentioned it because he assumed everyone knew. Nobody else knew. That single insight, once documented, reduced startup scrap by 12% when the operator was on leave.

Step 3: Write One-Page Process Sheets That Someone Will Actually Read

The output of your walk-and-talk sessions should not be a 40-page manual. Nobody on your floor will read a 40-page manual. The output should be a set of one-page process sheets—one per critical task or machine. Each sheet follows a strict format:

• Task name: What the sheet covers (e.g., “Press 3 Die-Set Changeover”).
• Who this is for: The operator or technician who will use it.
• Before you start: The 3–5 checks or conditions that must be met.
• Step-by-step: No more than 10 steps. Use short sentences. Use photos or hand-drawn sketches if they help.
• Watch for: The 2–3 things that commonly go wrong and how to catch them early.
• When to stop and call someone: The conditions under which the operator should not try to fix the problem alone.
• Last updated: Date and name of the person who validated the sheet.

Print these sheets. Laminate them if the environment allows. Keep them at the point of use—on the machine, at the workstation, in the dispatch area. The test of a good process sheet is whether a reasonably skilled worker who has never done this specific task can follow it and produce acceptable output on the first attempt. If they cannot, the sheet is not detailed enough.

This is not about creating a corporate knowledge base. It is about creating a survival document for your operations. Think of it as the factory equivalent of a production runbook, a concept borrowed from site reliability engineering, where structured documentation ensures someone can respond to a known situation without rediscovering the solution. Your process sheets are your production runbooks. They protect you when the person who wrote them is not there.

When you are turning raw walk-and-talk notes into these sheets, an AI story generator can help structure raw notes into clear steps, but the real test is whether a new operator can follow the sheet and produce good output on the first try.

Step 4: Build a Simple Knowledge Register

Once you have a stack of process sheets, you need a system to keep them alive. I recommend a physical binder—a three-ring file kept in the production office or the maintenance room. Call it the “Operations Knowledge Register.” It contains:

• A table of contents listing every process sheet, the machine or area it covers, and the date it was last reviewed.
• The process sheets themselves, organized by area (press shop, assembly, packing, maintenance, quality).
• A log page at the front where anyone can note a change, a correction, or a new problem that needs to be documented.

If your team is comfortable with digital tools, a shared folder on Google Drive or a WhatsApp group with pinned files works. But I have seen too many digital initiatives fail because the Wi-Fi was down or the login was forgotten. A physical binder is harder to ignore. It sits on the shelf and stares at you during the morning meeting.

The register is not a one-time project. It is a living document. Every time a machine is modified, a new customer specification arrives, or a workaround becomes permanent, the relevant sheet must be updated. Assign ownership: the area supervisor is responsible for keeping their section’s sheets current. Make it part of the monthly operations review to spot-check two or three sheets for accuracy. If a sheet is wrong, it is worse than no sheet at all because it teaches the wrong behavior.

Step 5: Make Documentation a Daily Discipline, Not a Project

The biggest mistake I see is treating tribal-knowledge capture as a one-time initiative. The owner gets worried, runs a documentation drive for two weeks, produces a binder, and then never touches it again. Two years later, the binder is outdated, and the new senior operator has developed a fresh set of undocumented tricks.

Documentation must become part of the operating rhythm. Here are three practices that work in small factories:

1. Five-minute debrief after any unplanned downtime. When a machine stops unexpectedly, the operator and supervisor spend five minutes writing down what happened, what fixed it, and what should be checked next time. This goes into the knowledge register as an addendum to the relevant process sheet.
2. “Teach-back” during training. When a senior worker trains a junior, the junior writes the process sheet as they learn it. The senior worker reviews and corrects it. This forces the senior worker to articulate what they know, and it produces a document that is written at the right level for the next learner.
3. Quarterly knowledge review. Every quarter, pick one critical area and spend two hours walking the floor with the person who knows it best. Ask: “What has changed since we last wrote this down?” Update the sheets.

These practices do not require new software, new hires, or a consultant. They require the owner or plant manager to insist that documentation is part of the job, not an extra task to be done when there is free time. There is never free time. You make time by deciding that preventing a three-shift downtime event is worth 45 minutes a week.

Why This Is an Operations Problem, Not an HR One

I want to return to the title of this article because the distinction matters. HR can help with exit interviews, succession planning, and training calendars. But HR does not know what a die-set changeover looks like. HR does not know which supplier’s raw material runs differently on Machine 2. HR cannot judge whether a process sheet is accurate enough to prevent scrap.

Tribal knowledge capture is an operations problem because it directly affects throughput, quality, and cost. It belongs in the same category as preventive maintenance, production scheduling, and quality inspection. The operations leader—whether that is the founder, the plant manager, or the production head—must own it. They must decide which knowledge is critical, allocate the time to capture it, and enforce the discipline to keep it current.

I have seen family businesses where the founder is the single point of failure for 40% of the operational decisions. The founder knows it. The family knows it. But nobody wants to have the conversation because it feels like planning for the founder’s absence, which feels disloyal. This is not disloyalty. It is operational risk management. If your factory cannot run for two weeks without you, you do not have a business. You have a job that you own.

What Happens When You Get This Right

Let me tell you about a fastener manufacturer in Faridabad I worked with three years ago. They had a senior thread-rolling operator named Gurmeet who was planning to retire in 18 months. Gurmeet knew the setup parameters for over 200 different SKUs. He carried them in a small notebook that he kept in his shirt pocket. Nobody else could set up the thread-rolling machines without calling him, even on his off days.

We spent four months doing structured walk-and-talk sessions with Gurmeet. We produced 23 process sheets covering every major setup category. We photographed his notebook pages and translated his handwritten notes into clear steps. We had a junior operator practice setups using only the process sheets, with Gurmeet watching and correcting the sheets where they were unclear. By the time Gurmeet retired, the junior operator could handle 80% of the setups independently. The remaining 20%—the rare, complex SKUs—were documented well enough that a supervisor could guide any operator through them.

The owner told me later that the process cost him about 120 hours of Gurmeet’s time and his own time combined. He calculated that a single day of downtime on the thread-rolling section would have cost him more than that investment. The documentation has since been updated twice as new SKUs were added. It is now part of the standard training for every new operator in that section.

That is the standard. Not a perfect knowledge base. Not an AI-powered documentation system. A set of one-page sheets that actually work, kept current by the people who use them, owned by the operations leader.

Getting Started Tomorrow Morning

If you are reading this and recognizing your own factory, here is what I suggest you do tomorrow morning:

1. Walk your floor and make a list of the five people whose absence would hurt you most within a week. Write their names and their critical knowledge areas on a whiteboard.
2. Pick one of them—ideally the one closest to retirement or the one whose knowledge is narrowest and deepest. Schedule a 45-minute walk-and-talk session with them this week.
3. During that session, ask the five questions I listed earlier. Record the answers. Write a one-page process sheet from those answers before the end of the week.
4. Test the sheet. Have someone else try to perform the task using only the sheet. Fix what is missing.
5. Put the sheet in a binder. Label the binder “Operations Knowledge Register.” Tell your supervisors that this binder is now part of how the factory runs.

That is it. No software purchase. No consultant engagement. No six-month project plan. Just the first sheet, tested and filed, with a commitment to do one more next week.

The cost of not doing this is not theoretical. It is measured in downtime hours, scrap kilograms, and missed delivery deadlines. Prakash’s corrugator was down for three shifts. Your machine might be next.