Process Debt: The Hidden Tax of Undocumented Workarounds

A shipment needs a customs code that is not in the system, so an operations manager picks the one she has used for years and it clears. A pricing exception for a strategic customer lives in a rule only the head of sales remembers to apply. A month-end reconciliation only balances because someone runs a spreadsheet macro nobody else has opened. None of this is written down anywhere. It works, quietly, for years. Then that person leaves, and three processes that looked healthy on the org chart stall at once, because the reasoning behind every shortcut walked out with them.

This is process debt, and it is one of the most expensive liabilities in a mid-sized company precisely because it never appears on a balance sheet. Every undocumented workaround, every just-ask-Maria shortcut, every off-book exception is a small loan the company took out to move faster today. Like any debt it charges interest, paid in the extra time, errors, and interruptions every future colleague spends, and it compounds silently as the workarounds pile up. The metaphor is borrowed on purpose: in 1992 Ward Cunningham described exactly this dynamic for software and called it technical debt¹. Process debt is the same mechanism applied to how people work, and it is larger, because there is no codebase to scan.

The reflex fix is to demand more documentation, run another handover process, or buy another wiki. None of it pays the debt down for long, because you are fighting a structural problem with willpower. This guide is for the Geschaeftsfuehrer, operations lead, or IT director who is tired of watching a single departure freeze the business, and wants the mechanism that actually pays process debt down: a Company Brain, a living memory fed by daily work and feedback that survives turnover, with AI employees running the routine work on top of it across email, Teams, SharePoint, CRM, and ERP.

What Process Debt Actually Is

Process debt is not disorganisation or a messy shared drive. It is a specific liability: the accumulated gap between how a process is supposed to work and how it actually works, held together by shortcuts that live in people rather than in any system. The name is deliberate. Ward Cunningham coined technical debt to explain to non-technical stakeholders why shortcuts in code carry a future cost, and the same logic transfers cleanly to process^1,3.

• It is a loan, not a mistake - a workaround is often the right call in the moment, exactly as a quick code fix can be; the debt is the unpaid obligation to capture how and why it works¹.
• It charges interest - every future colleague who has to ask, re-derive, or guess pays the interest, and Martin Fowler notes the whole value of the metaphor is comparing that interest against the cost of paying down the principal².
• It compounds - workarounds stack on workarounds, exceptions get exceptions, and the interest rate rises as the tangle grows and fewer people understand it.
• It is invisible - unlike technical debt, which at least sits in a codebase engineers can read, process debt hides in inboxes, chat threads, and the heads of long-serving staff, so no one sees the balance⁶.
• It comes due at departure - the principal falls due the day the person who held the workaround leaves, because the reasoning was never written down.

The classic articulation of the mechanism came from the person who named it, describing code, but the words fit process exactly.

Swap not-quite-right code for not-quite-documented process and the sentence describes the average Mittelstand back office. The debt speeds things up, nobody pays it back, and the interest accrues until the day it does not.

For the specific case where a single person becomes the concentration point for that risk, our piece on the bus factor and how one person leaving stalls the whole company covers the key-person angle in depth.

The Four Ways Process Debt Accrues

Process debt does not build up in one way. It accrues down four separate channels, and any one of them is enough to freeze a process when the wrong person is out. Most documentation efforts patch only the first, which is why the balance keeps growing.

Channel 1: The undocumented workaround

• The system does not fit reality - a field is missing, a rule is wrong, or the tool cannot do what the situation needs, so someone invents a manual patch.
• The patch becomes permanent - the workaround works, so it is never revisited, and the temporary fix quietly becomes load-bearing infrastructure.
• The reasoning is never captured - why the workaround exists and when it applies stays in the head of whoever built it, so the next person cannot judge it.

Channel 2: The just-ask-Maria dependency

• One person becomes the answer - instead of a documented process, the company has a person, and everyone routes questions to them⁶.
• Most role knowledge sits in one head - Panopto puts it at 42 percent of the knowledge people need to do the job, unique to the individual holding it⁸.
• The dependency feels efficient - asking Maria is faster than finding a document, so the debt is convenient right up until Maria leaves.

Channel 3: The off-book exception

• The real rule is not the written rule - a customer gets special pricing, a supplier gets a different approval path, a case skips a step, and none of it is in the official process.
• Exceptions breed exceptions - once one case is handled off-book, others follow, and the shadow process grows larger than the documented one.
• Nobody can see the full set - because the exceptions live in memory and email, no one holds the complete list, so they cannot be audited or transferred.

Channel 4: The tribal shortcut

• Experience compresses the process - a veteran skips three steps because they know they are unnecessary here, and the shortcut is faster and correct, but unwritten⁷.
• The shortcut looks like the process - newcomers copy the visible behaviour without the judgement behind it, and get it wrong on the cases where the shortcut does not apply.
• Retirement drains it fastest - KfW warns that the retiring boomer generation is taking established routines and experiential knowledge with it, most of which was never documented¹⁶.

For the retirement-driven version of this, where decades of tribal shortcuts leave at once, see our piece on the retirement knowledge cliff facing the Mittelstand.

What Process Debt Actually Costs

The cost of process debt hides because it never arrives as an invoice. It is paid in interest, thousands of small delays and re-solves, and then in one large principal payment when a concentration-point person leaves. Add it up and it is one of the largest unmanaged costs in a mid-sized company.

• Most role knowledge is stranded in one head - Panopto found 42 percent of the knowledge employees need is unique to the individual who holds it, so every workaround is a single point of failure⁸.
• Hours a week go to waiting for the person who knows - employees lose around five hours a week waiting to reach the colleague with the unique knowledge, and one in ten wait twice that⁹.
• More hours go to searching and redoing - Panopto puts time lost searching for information at roughly eight hours a week and time redoing work already done elsewhere at about six hours⁹.
• Frustration is near-universal - 81 percent of employees report frustration when they cannot get the information they need to do their job properly⁸.
• Poor knowledge sharing has a price tag - IDC and Panopto research puts the cost of inefficient knowledge sharing at roughly 4.5 million dollars a year per 1,000 employees, and around 47 million for a large enterprise¹⁰.
• Departure calls in the principal - Gallup puts the cost of replacing an employee at one-half to two times their annual salary, and for a process-debt concentration point the true cost runs higher because their exit stalls the work only they understood¹¹.

A worked model makes the leak concrete. Consider a 500-person mid-sized company, and treat these as overlapping estimates rather than figures to be summed.

Our detailed breakdown of what having no Company Brain really costs works these numbers through in full.

Why Process Debt Comes Due in 2026

Undocumented workarounds are not new, but several forces converged to turn a slow-burning cost into an acute one. The debt that companies carried quietly for years is being called in.

1. The demographic wave hit - as the boomer generation retires, decades of tribal shortcuts leave at once, and KfW research names the loss of undocumented routines and experiential knowledge as a direct consequence¹⁶.
2. The talent gap makes replacement slow - Bitkom counts more than 100,000 unfilled IT roles in Germany, so a departure that calls in process debt cannot be back-filled quickly¹⁵.
3. Work fragmented across more tools - process knowledge now lives in email, Teams, SharePoint, CRM, ERP, and chat at once, so the workarounds are scattered and even harder to see or capture.
4. AI made undocumented process dangerous - point an AI assistant at your idealised documentation and it will confidently give the answer the process does not actually follow, so the debt now causes bad decisions at scale, not just slow ones.
5. Agent projects are failing on it - Gartner predicts over 40 percent of agentic AI projects will be cancelled by the end of 2027, often because the agents lack the real process context, and Gartner now warns of AI debt accruing on top of the debt companies already carry^13,14.

The strategic point holds whatever stack you run: the future of company process is a living memory that captures how work is really done, not a folder of documents that says how it is supposed to be done.

A Worked Example: The Exception That Came Due

Abstractions hide the cost, so here is a concrete, composite example from the kind of mid-sized manufacturer Superkind works with. Follow one undocumented workaround from convenient shortcut to frozen process.

One workaround, one departure, three frozen weeks

1. Year one: the loan - a strategic customer needs a non-standard delivery and invoicing split that the ERP cannot model cleanly. The order manager builds a manual workaround in a spreadsheet plus a set of unwritten rules, and it works flawlessly for years.
2. The interest, paid quietly - every time a colleague covers for her, they interrupt her to ask how the split works, or get it slightly wrong and trigger a credit note. Nobody logs any of this as a cost.
3. The principal falls due - she leaves for a competitor. The next month-end, the split breaks, invoices go out wrong, the customer escalates, and three people spend three weeks reverse-engineering a process one person ran in an afternoon.
4. The root cause - no one acted in bad faith. The company simply never captured the workaround or its reasoning, so a convenient loan became an emergency the day it was called in.

The reason the memory could keep running the process is that it captured the reasoning, not just the steps. Our piece on capturing the why behind decisions, not just the what explains why the reasoning is the part worth keeping.

Why Static Wikis and SharePoint Never Pay It Down

The instinct is to pay down process debt with a better repository: a wiki, a SharePoint site, a stack of SOPs. But a repository is passive, and passive stores capture outcomes, not the reasoning and exception-handling that make up the debt. The difference between a document store and a Company Brain is not a feature list; it is a difference in what the thing fundamentally is.

The core distinction

• A wiki captures the outcome - it records the idealised final process, not the workarounds and exceptions that actually run it, so the debt stays invisible⁶.
• A Company Brain captures the real process - it is written to by the ongoing work, so it reflects the workaround and the reasoning, not the poster on the wall.
• A wiki is read; a memory is used - people consult a wiki occasionally, but AI employees use the Company Brain on every task, which keeps it exercised and current.
• A wiki decays; a memory learns - a page goes stale the moment the process changes, while a Company Brain updates itself from the next piece of work and the next correction.
• A wiki is a silo; a memory is connected - it draws from email, Teams, SharePoint, CRM, and ERP as live sources rather than a snapshot someone pasted in months ago.

For the narrower question of why SharePoint specifically cannot serve as a living knowledge base, our companion piece on SharePoint as a knowledge base covers the limits in detail.

How a Company Brain Pays Down Process Debt

A Company Brain does not pay down process debt by being a bigger, better-organised wiki. It pays it down by changing what captures the process in the first place. Three mechanisms do the work, and all three are automatic side effects of using it, not extra chores.

The three debt-paying mechanisms

1. It observes the work - because AI employees perform the routine tasks, the real process, including the workaround and its reasoning, is captured as a by-product, so there is no documentation step to skip.
2. It learns from feedback - every correction a person makes updates the shared memory, so an exception handled once is not re-derived and a mistake fixed once does not recur.
3. It reads from live systems - prices, stock, approvals, and records come from the source of truth in your CRM, ERP, and files, not a rule someone remembered to apply.

The learning mechanics matter enough to stand alone. Our deep dive on the feedback loop that makes AI employees better every week covers exactly how a correction becomes retained process knowledge.

The Pay-Down-the-Process-Debt Playbook

You do not retire process debt with a company-wide documentation programme; you start in one function where the debt is heaviest and pay it down there first. Here is a practical 90-day sequence.

Phase 1: Find the debt and connect the systems (Weeks 1-4)

1. Week 1: Find your concentration points - identify the functions where work stalls when one person is out, such as order intake, month-end close, quoting, or internal IT support.
2. Week 2: Map the real process, not the poster - sit with the people who hold the workarounds and trace how the work actually runs, exceptions and all, so you know the true balance.
3. Week 3: Connect an AI employee to the live sources - email, Teams, SharePoint, CRM, and ERP, so answers come from the source of truth rather than a remembered rule.
4. Week 4: Seed the memory and set feedback rules - load the good existing knowledge as context and agree how corrections are captured, so the loop works from day one.

Phase 2: Capture the real process in the work (Weeks 5-8)

5. Week 5-6: Run in parallel with the team - the AI employee handles routine work alongside people, who correct it, and the memory captures the workarounds and exceptions as they are used.
6. Week 7: Log the interest you stop paying - count the cases where the memory answered instead of interrupting the expert, because that is process debt being paid down in real time.
7. Week 8: Resolve contradictions and retire stale pages - where the documented process and the real one conflict, settle it once and point people to the living memory.

Phase 3: Make it the source of truth and expand (Weeks 9-12)

9. Week 9: Shift the source of truth - point the team to the Company Brain for how the process actually runs and keep old documents only for archival reference.
10. Week 10-11: Capture the exceptions in use - as edge cases arise they are handled and fed back, so the hardest, most-undocumented part of the debt gets paid down where it is used.
11. Week 12: Report and pick the next function - present the interest saved and the departures de-risked, then repeat the cycle in the next debt-heavy department.

The change-management side matters as much as the technical side. Our guide on onboarding your team when AI employees join covers making the shift stick.

How Superkind Fits

Superkind builds AI employees for the Mittelstand that carry routine work and, in doing so, build a Company Brain: a living memory of how your company actually operates, workarounds and all. The point is not a shinier place to store SOPs. It is to pay down process debt, so a single departure stops freezing the business.

• Company Brain, not a document store - the memory is built and used by AI employees every day, so it captures the real process rather than an idealised outcome.
• Connects to your existing systems - email, Teams, SharePoint, CRM, and ERP feed one live memory instead of a dozen scattered silos, with no rip-and-replace.
• Captures the reasoning, not just the steps - the why behind a workaround gets recorded, so it can be judged, transferred, and trusted.
• Learns from daily feedback - every correction updates the memory, so an exception handled once is not re-derived and accuracy compounds week over week.
• Holds the exceptions - the off-book cases get captured as they are handled, before the person who knows them leaves.
• Survives turnover - because the process was captured in use, a departure no longer calls in the principal all at once.
• Process-first discovery - we map how your team actually works, exceptions included, before building, so the memory fits your workflows rather than a generic template.
• Compliant by design - data stays in your infrastructure, access is controlled, and the memory is observable for DSGVO and EU AI Act record-keeping.

To see how the same memory stays under your control as it grows, read our piece on a Company Brain that stays sovereign.

Decision Framework: How Much Process Debt Are You Carrying?

Not every company needs to act tomorrow. Use these signals to judge how much process debt you already carry and where to start paying it down.

The goal is old and the goal is right: know how your work actually gets done. What changed is the mechanism, from a store people must maintain to a memory the work maintains for them.

Frequently Asked Questions

Henri Jung

Co-founder of Superkind, where he helps SMEs and enterprises deploy custom AI employees that actually fit how their teams work. Henri is passionate about closing the gap between what AI can do and the value it creates in real companies. He believes the Mittelstand has everything it needs to lead in AI - it just needs the right approach: a company that captures how its work really gets done instead of paying interest on undocumented workarounds forever.