Three clients. Three different businesses. The same failure.
Each one had a thing that worked well. Each one had that thing owned entirely by a single person. And each one eventually faced the moment when that person was unavailable, departed, overwhelmed, or simply not around when the process needed to run.
The thing stopped working.
The influencer program at Drups Ventures
One of the brands in Joseph's portfolio had an influencer program. It had been running, generating content, driving some sales. Then it stopped. Not because the influencers disappeared or the strategy failed. It stopped because the person running it was no longer running it, and nobody had written down how it worked.
When the engagement picked it up, the program existed in name only. There was no roster of active influencers. No documented outreach process. No content submission workflow. No clear rules on what got approved and what didn't. Payment was through Refersion at a $100 threshold, but that was in Refersion's system, not in any document the team could reference.
Rebuilding the program meant building it from scratch in Asana: outreach steps, approval criteria, content submission requirements, payment triggers. The goal was to have a new campaign running by October 1st. Whether that deadline was hit or not, the rebuild took weeks of work that should not have been necessary. The program had existed before. The knowledge had walked out with the person who ran it.
The fix is straightforward. An influencer program needs, at minimum: a roster with contact information and last activity date, an outreach template library, documented approval criteria, a content submission and review workflow, and payment processing rules. If those five things exist in writing, the program survives personnel changes. If they exist only in someone's head, they don't.
The supplier relationships at Realsy
Realsy's supply chain runs through a date farm in Mexico and Camel Foods for nut butters. The relationships with those suppliers had depth: established contacts, agreed payment structures, quality standards, and a history of how problems got resolved when they came up.
That context lived primarily with Austin, the founder. When the operations team, including the consultants on the engagement, needed to interact with suppliers directly, the knowledge wasn't documented. Supplier contact information had to be tracked down. Payment terms had to be confirmed in real time rather than referenced from a source of record. When a production delay or a quality issue came up, the team didn't have a documented escalation path. They had to ask Austin.
Austin was sometimes available and sometimes not. When he wasn't, things stalled.
The specific risk this creates: a supplier relationship with no documentation is a supplier relationship that breaks down the moment the person who owns it is sick, on vacation, or leaves the company. The supplier doesn't know they should call someone else. The operations team doesn't know who to call on the supplier side or what was agreed in previous conversations.
Minimum viable documentation for a supplier relationship: primary contact and backup contact, payment terms per PO structure, quality acceptance criteria, lead time ranges including historical variance, and a record of any standing agreements or accommodations that exist outside the standard contract.
The forecasting model at Samako
At Samako, the master forecasting sheet was built by the consultant. It covered stock coverage calculations, PO tracking, NSPA container management, and dual-mode forecast switching between top-down and operational views. It was designed to reduce the time spent on the monthly rolling forecast from over 5 hours to something manageable.
The problem was that only the consultant fully understood it.
Boet, the owner, could use the inputs and outputs. He could enter payment confirmations and read the coverage numbers. But the logic underneath the model, the formula structure, the forecast mode switching, the reorder date calculations, and the reason for each column's existence, was not documented. The model worked because the consultant was maintaining it. If the consultant became unavailable, the model would gradually decay as small errors accumulated without anyone knowing how to diagnose them.
This is a specific version of the single-person dependency problem that shows up in consulting engagements constantly. The consultant builds something useful. The client uses it. The knowledge of how it works stays with the consultant. The client has an output, not a capability.
Making the Samako model transferable required writing explicit documentation for each section: what the formula is calculating, what inputs feed it, what to check when a number looks wrong, and how to update the model when new SKUs or suppliers are added. Color-coded status indicators on the process flow charts helped, but they supplemented documentation rather than replacing it.
What makes a process real
A process is real when a person who has never run it before can pick up the documentation and execute it without asking anyone for help.
That's the standard. It's a high bar, and most operational processes don't meet it. But the bar is right. Anything less is a dependency risk waiting to become an operational crisis.
The minimum viable SOP for any repeatable process has five elements: the trigger (what starts it), the steps in order, the decision criteria for anything that branches, the output (what done looks like), and the escalation path (who to call when something breaks that isn't covered by the steps).
One page is often enough. Processes that require 20 pages of documentation usually haven't been simplified enough to be executable. The goal isn't comprehensive coverage of every edge case. It's enough structure that the process doesn't stop when the person who designed it isn't there.
The test is simple: hand the SOP to someone who has never done the task and watch what happens. If they can complete it without asking questions, the SOP is real. If they get stuck, you found the gap. Fix the documentation, not the person.
