The Human in the SOP
Be the boots on the ground
Be the boots on the ground
I am not anti-tech. I build with it, rely on it, live in it. But when I reach for a search bar or an AI window, I skip the part where a person shows me how it actually works. That skip looks harmless. It is not. It shows up later when something fails and the answer I grabbed was clean but thin.
I have written hundreds of SOPs over twenty years. Warehouse receiving. Medication handling. Operating finicky lab equipment that behaves like a different machine when the room is a few degrees warmer. The through-line in all of it is simple. Great SOPs start as human conversations. Bad SOPs start as exports.
What gets lost when we skip people
Search gives answers. People give context. Search collapses time. People expand it. In the compression, you lose the slow, earned parts of knowledge that make procedures durable.
When someone teaches you in person, you do not just get steps. You get the pace, the pressure, the “stop here” warning that never shows up in a tidy checklist. You hear why the shortcut fails on rainy days. You see the pause before step three because that is where people ruin it. Those micro-signals are the difference between compliance on paper and performance on the floor.
Why this matters to SOPs and the business
SOPs are not documents. They are operational contracts. They turn tribal knowledge into repeatable action. They cut onboarding time. They reduce variance. They make audits boring. Done right, they protect people and revenue at the same time.
The fast-answer trap
Instant answers feel great. The problem is that speed wears the mask of mastery. You copy a clean explanation, run it once, and your brain files it under known. It is not known. It is rehearsed text. Without friction, there is no feedback. Without feedback, you do not correct. That is how fast knowledge feeds false confidence, which becomes a fragile SOP, which becomes downtime, scrap, a CAPA you did not want, and an ulcer.
How knowledge actually grows into an SOP
We like to think the right answer is fixed. In practice, the right answer gets bent by constraints until it fits the job in front of you. That bending looks like manipulation. It is actually progress. Someone takes a method, trims it for nights, adds a step for low staffing, removes a step for mornings, and hands you a version that works here. Then you hand a better version to the next person. That is how knowledge compounds. SOPs should capture that evolution, not freeze the first draft.
What belongs in a real SOP that a search result will never give you
Boundary conditions. Temperature, humidity, torque, hold times. The edges where success flips to failure.
Cues and tells. The whine of a rotor that means stop. The sheen on a surface that means go. The smell that means throw it out.
Escalation rules. Who to call, by role, with numbers that work. What to do while you wait.
Common failure modes. What goes wrong after step two and how to catch it early.
Adjustments under constraint. How to run it with half the staff, or during a power dip, or when the vendor switches lot numbers.
Records that matter. The one log auditors always ask for first. Where it lives. How to reconcile it when two systems disagree.
The why. One short line that explains the risk the step is controlling. Adults follow rules better when they understand the risk.
A short story from the floor
I once documented a centrifuge procedure that had passed every paper review and still wrecked samples on Tuesdays. The fix did not come from a manual. It came from an operator who said it only happens when the morning deliveries sit in the hallway for a while. The hallway had a heater grate. The samples took a small temperature bump before processing. We added a 10-minute temperature equilibration step to the SOP and a line to relocate drop-off bins. Problem ended. Nobody would have found that in a forum thread.
My playbook for SOPs that survive contact with reality
Sit with the best operator and the newest one. The best shows craft. The newest shows what the document hides.
Watch the work twice. First for flow. Second for the stumbles.
Ask five questions that expose the edges:
What usually goes wrong right after step two, and how do you catch it early?
2. If you had to do this with half the time or half the staff, what would you change first?
3. What did you try that looked good on paper and failed in practice?
4. Show me once, then watch me do it, and tell me what I am missing.
5. What would you never do again, even if someone swears it saves time?
Draft the SOP in plain language. Verbs up front. One action per line. No fluff.
Build in checks. Acceptance criteria, go/no-go points, and visible cues. Pictures beat paragraphs.
Pilot it with real people. Not the author. Run it at the worst time of day.
Measure it. Time to complete, error rate, rework, training time. If it does not move a number, it is theater.
Train with demonstration and return demonstration. If they cannot teach it back, they do not have it yet.
Version control like it matters. It does. Date, owner, change reason, redlines.
Audit the reality. Unannounced walk-throughs. Does the work match the words. Fix the words if the work is right. Fix the work if the words are right.
Where AI belongs in this process
Use AI to accelerate the grunt work. Structure, format, cross-references, regulatory citations, table templates, quizzes, translation for multilingual teams. Use AI to draft a first outline or to turn bullets into clean prose. Do not let AI invent the parts that come from scar tissue. The nuance comes from people who have failed and learned and are willing to tell you about it.
The cost of getting this wrong
Thin SOPs look efficient. They are not. They create training debt and hidden risk. They inflate near misses. They turn audits into rescue missions. They chew up managers who spend their day clarifying what the document should have settled. The bill comes due in injuries, write-ups, lost product, and lost trust.
The upside of getting it right
Thick with the right detail does not mean thick with words. It means the document carries the weight of experience in a way a new hire can lift. It means you can move people across sites without starting from zero. It means less variance, fewer surprises, cleaner CAPAs, smoother handoffs, and a culture that treats procedures as tools, not obstacles.
The trade I am willing to make
Choosing a person over a search bar is a trade. You give up speed. You gain depth, accountability, and a version of the answer that fits your world. You also gain a name and a story attached to the skill. That makes the knowledge stick. It makes it yours. Then you lock it into an SOP so the next person does not have to guess.
Use search. Use AI. Then ask a human. Walk the floor. Listen for the part that never shows up online. Write that into the SOP. Teach it. Audit it. Iterate it. That is how you turn answers into competence, competence into trust, and trust into a business that runs when you are not watching.