Undocumented processes are the cheapest thing in the business, until they are the most expensive thing in the business.
They feel free. There is no invoice. No subscription. No project on the roadmap. Sarah knows how to do it, and when someone new joins, Sarah shows them. The cost does not show up on any line in the P&L, which is why it is almost always under-counted.
When someone eventually tries to build the business case for documentation - SOPs, work instructions, a proper knowledge system - they often struggle, because the status quo looks cost-free. It is not. It is just distributed across four lines that no one adds up.
Here is how to add them up.
Cost line 1: rework and errors
When a process lives in someone's head, it tends to be executed with drift. Not because the person is careless, but because humans interpret instructions differently each time they give them, and interpret them differently again when they receive them.
The drift shows up as rework. A job done to the wrong spec that needs redoing. A defect that makes it to the customer. A step skipped that someone else has to go back and complete. A hand-off where the next person cannot tell what state the work is in.
To estimate this conservatively, ask three questions for a given process.
- How often does this process run? Weekly, daily, per-shift, per-customer?
- What is the error rate? Not the reported rate - the honest rate. If you do not know, shadowing three or four runs usually yields a defensible estimate.
- What does each error cost to fix? Include the time of the person who fixes it, plus any material or customer-impact cost.
Multiply and you have an annualised rework cost for that one process. Most teams do this for their largest two or three processes and find the number larger than they expected. The point is not to be precise; it is to be conservative and still end up with a number that justifies attention.
Cost line 2: slower onboarding
New starters ramp at the speed of available information. If every answer requires asking a colleague, onboarding runs at the speed of colleague availability - which is to say, slow, with gaps whenever the colleague is on leave or deep in their own work.
To cost this line, you need three numbers.
- Time to productive contribution.How many days from start until the new starter can do the role's core tasks without asking for help.
- Loaded daily cost. Salary plus benefits plus employer taxes divided by working days. Most finance teams have this.
- How much faster onboarding could be with proper documentation. A conservative estimate is a reduction of a third for most roles; more for tightly procedural roles.
Multiply loaded daily cost by the number of days saved, then by hires per year. That is the onboarding line. Do not forget to also count the time of the colleagues who answer the new starter's questions - often a quarter or more of a senior person's day, for weeks.
CIPD's surveys on onboarding consistently show that well-onboarded staff both ramp faster and stay longer. The retention effect is a separate cost line, which most business cases under-count because the replacement cost is paid in a different quarter than the savings.
Cost line 3: risk exposure
This one is the most uncomfortable to estimate, which is why it gets skipped. The cost of undocumented process shows up as risk exposure: the gap between what you tell a regulator, insurer, auditor, or customer, and what you can actually prove.
“Yes, we train everyone on this.” Prove it. “Yes, we follow the proper procedure.” Prove it. “Yes, that member of staff was competent.” Prove it.
Most teams can answer these truthfully in the affirmative. They just cannot prove it in under an hour with a shared folder full of Word documents and a couple of email threads. The cost of not being able to prove it shows up in three places.
- Audit time. Hours - sometimes weeks - of senior staff time spent assembling evidence that would have assembled itself if the system had been collecting it all along. The HSE and sector regulators typically make clear what kinds of records they expect; retrieving them should be a query, not a project.
- Penalties and remediation. When the evidence does not exist, the outcome is often a remediation plan with a deadline and, sometimes, a financial penalty. The penalty is visible. The remediation cost usually is not.
- Insurance and contract terms.Insurers and enterprise customers increasingly ask about documented processes and training records as part of due diligence. “We have it but it would take a week to get to you” is read as “we do not have it.”
Pricing this line is hard because the cost is probabilistic. A reasonable approach is to estimate the worst-case cost of one moderate incident in your industry - a safeguarding failure, a data incident, a major customer escalation - and multiply by a conservative probability. You will end up with a number you would pay real money to avoid.
Cost line 4: key-person dependency
The fourth cost is the one that tends to surface dramatically rather than gradually. Sarah knows how the invoicing process works. Tom is the only one who understands the supplier ordering system. Priya is the de facto expert on the bespoke workflow that half the revenue flows through.
When one of these people takes annual leave, the process slows. When one of them is off sick, bottlenecks form. When one of them leaves the company, real revenue is at risk, because the knowledge walks out with them and nobody has a document that lets the next person step in.
To estimate this, do not try to price “if Sarah left.” Instead, price the ongoing drag of the dependency itself.
- How much of Sarah's time is spent answering questions nobody else can answer? That is time she cannot spend on higher-value work.
- How often is a process delayed because the key person is unavailable? Delay-hours add up across a year.
- What is the estimated cost to the business if a specific key person left unexpectedly, including hiring, ramping, and the disruption in between?
Multiply the third bullet by an honest annual probability of departure and you have a defensible figure for the ongoing cost of the dependency. For roles with specialised knowledge, that number is usually significant enough to fund documentation several times over.
A worked example: a 25-person services firm
Suppose you run a twenty-five-person services business. You do not have formal SOPs. Work gets done by a mixture of email, Slack, shared docs, and lived experience. Three or four senior staff hold most of the operational knowledge. You hire three to five people a year.
Rough, conservative estimates you might build on a whiteboard:
- Rework and errors. Across your two highest-volume processes, perhaps several hundred hours a year of rework at a loaded hourly cost. Tens of thousands of pounds, easily.
- Slower onboarding. If proper documentation saved each new starter three weeks to full productivity, at loaded cost, across four hires, you are into another mid-five-figure line.
- Risk exposure. One moderate compliance issue triggered by an absent record can cost more than the entire documentation programme. Most teams price this as a low-probability, high-impact reserve and still end up with a figure that justifies investment.
- Key-person dependency. If one of your three senior staff left unexpectedly, the immediate cost - hiring, ramping, disrupted client work - is often in the tens of thousands before you count lost revenue. Even a five or ten percent annual probability makes the ongoing expected cost substantial.
Add them up honestly and you are looking at a cost of going undocumented that is almost certainly six figures a year for a business of this size. The documentation programme to eliminate most of it - a proper SOP library, structured onboarding, training records with evidence - is a small fraction of that.
Why these numbers feel big. Because each line is plausible on its own, and nobody ever adds them up. The CFO sees hiring cost on one line, insurance on another, rework on a third - and none of them gets traced back to the same root cause.
How to use this calculation
The point of costing undocumented process is not to produce a spreadsheet that survives cross-examination. It is to shift the internal conversation from “documentation is nice to have” to “the status quo is already expensive, and the money is leaving through leaks we can name.”
Three suggestions for using the numbers well.
- Be conservative. Use the low end of each range. The credibility of the case depends on the numbers feeling defensible, not impressive.
- Tie each line to a specific process.“We are losing roughly X on rework of the customer onboarding process” is harder to dismiss than “rework is expensive.”
- Make the counter-proposal specific.Not “we should document things.” A named set of priority SOPs, a named owner, a realistic timeline, and an explicit claim about which cost lines you expect to reduce and by how much.
The silent cost of undocumented processes is real, it is substantial, and it is almost always larger than the cost of fixing it. The first step is deciding to count. See how teams estimate the return on documented process if you want a starting point for your own calculation.