Most owners can name their biggest operational risk in about four seconds. It's a person. It's the controller who is the only one who knows how the month-end close actually works. It's the ops lead who remembers which customers get the custom pricing and why. It's the founder who still approves every quote because nobody else knows where the floor is.
Nobody wrote that risk down. It doesn't show up on the org chart, the balance sheet, or the systems diagram. It shows up the week that person takes a real vacation, and it shows up permanently the week they leave.
The fix starts with a boring exercise most companies skip: making a list.
Tribal Knowledge Is a Systems Problem Wearing a People Costume
When only one person can do something, the usual diagnosis is "we need better documentation." That's half right and mostly unhelpful. Documentation decays. People write an SOP, save it to a shared drive, and it's wrong within two quarters because the process changed and the doc didn't.
The deeper issue is that the process lives nowhere. It isn't in a system that enforces it, routes it, or logs it. It exists as a sequence of judgment calls in one person's head, stitched together with spreadsheets and email threads. That's not a training gap. That's missing infrastructure.
So before you assign anyone to "write it all down," find out what "it all" actually is.
Run a One-Week Inventory
For one week, have each person on your team keep a simple running log of every recurring task they touch — not projects, recurring work. Daily, weekly, monthly, quarterly. No formatting, no polish. Just a list.
Then add two columns to every line: who else can do this today, and where does this live. "Lives" means the actual home of the process: an ERP workflow, a CRM automation, a spreadsheet on someone's desktop, a doc, or nowhere at all.
You'll get a messy list. That's the point. Sort it and the pattern appears fast. A handful of processes will have exactly one name in the second column and "nowhere" in the third. Those are your two-person rule violations — the things that only one human on the planet can currently do for your business.
Grade the Risk by Consequence, Not Frequency
Not every single-owner task deserves the same urgency. Sort your list by what actually happens if that person is unavailable for two weeks:
- Revenue stops — quoting, order entry, fulfillment approvals, anything a customer is waiting on
- Cash stops — invoicing, collections, payroll, bank reconciliation
- Compliance slips — tax filings, certifications, regulatory reporting, access reviews
- Nothing visible happens for a while, then it's a mess — data hygiene, reconciliations, vendor management
That last category is the sneaky one. It creates no pain in week one and an enormous cleanup in month three. Give it real weight.
Fix It in the Right Order
There's a hierarchy here, and most companies reach for the weakest tool first.
Best: encode it in a system. If approvals, routing, pricing rules, and status changes live in your ERP or CRM, the process can't walk out the door. A workflow in Odoo that routes a quote above a threshold to a second approver doesn't need to be remembered. It just happens. Same with automated invoicing rules, reorder points, and hand-offs between sales and operations. When the system does the remembering, the human only has to supply judgment on the exceptions.
Second: build a real checklist tied to the system. Some work genuinely requires a person walking through steps. Fine — but the checklist should reference the actual screens, reports, and fields they'll use, and it should live where the work happens, not in a folder nobody opens.
Last: the long-form document. Useful for context and rationale. Terrible as the only safeguard.
Test It by Absence
A process isn't covered because someone says it's covered. Pick one high-risk item and run a dry run: the owner of that task goes hands-off for a full cycle while a backup runs it, with the owner available only for questions afterward. Log every question that came up. Those questions are the gaps in your system, stated precisely.
Do this a few times and something else happens. The people carrying those processes stop being bottlenecks and start being the ones designing the systems that replace the bottleneck. That's a promotion, not a threat — and framing it that way is most of the change-management work.
Why This Compounds
A business where critical processes live in systems rather than people is easier to scale, easier to delegate, easier to hand to a general manager, and materially more valuable to a buyer. Diligence teams ask a version of this exact question: what happens here if key people leave? You want a better answer than optimism.
If your inventory turns up more single-owner processes than you expected — and it usually does — that's the starting point for a real conversation about which ones belong in your ERP, which belong in an integration, and which just need a better checklist. Reach out to Infraxio and let's walk your list together.