Ask a site manager how the facility runs and you will get a confident answer. Ask them to leave for three weeks and see what happens. On most small and mid-sized mining operations, those two things are not consistent, and the gap between them is a risk nobody has priced.
Our guide to staffing a mining operation covers hiring, training and retention, and our overview of facility staffing roles covers how to structure the team. This is about the thing that is not in either document: what one person knows that nobody else does, and what happens to the site when they stop working there.
It is not the procedures that go missing
Operators who worry about this usually respond by writing procedures. That is worth doing, and it is not the thing at risk.
Procedures capture the work that is already understood well enough to describe. What leaves with a person is the layer underneath: which rack position keeps throwing errors regardless of what you put in it, which breaker has always run warm, what the hall sounds like when a fan bank is about to fail, which supplier actually answers the phone, and why a change made eighteen months ago was done the way it was rather than the obvious way.
None of that is in a procedure because none of it reads as knowledge while you hold it. It reads as familiarity. The person with it does not experience themselves as a risk; they experience themselves as someone who knows their site.
What the loss actually looks like
The failure is rarely dramatic on day one. The replacement is competent, the fleet keeps hashing, and the problem surfaces as a series of small, unexplained costs.
- Diagnosis slows down. Faults that used to be recognized instantly become investigations. The outage response that took forty minutes now takes a shift, because the first hour is spent rediscovering something the previous person simply knew.
- Known-bad positions get refilled. A slot that everyone quietly avoided goes back into rotation, and the machine in it starts underperforming. Nobody records it, because nobody knows it is a known problem.
- Relationships reset. The electrical contractor who understands your site, the utility account representative, the person at the repair depot who flags your units: those are personal, and they do not transfer with a job title.
- Decisions get re-litigated. Somebody proposes the obvious improvement that was already tried and abandoned for a good reason nobody wrote down, and the operation pays to learn it twice.
- Risk tolerance drifts. The new person does not know which shortcuts the site has learned are genuinely safe and which are not, so they either become too cautious to be effective or too casual to be safe.
This is adjacent to the concentration problem described in our piece on managing crew fatigue, and usually it is the same individual. The person who answers every night call is the person accumulating all the undocumented knowledge, which means the operation is carrying both risks in one body.
Find out what is concentrated before you try to fix it
The useful first step is not writing documentation. It is establishing what only one person can currently do, which most operators have never measured.
A short exercise produces most of the answer. List the tasks that have actually happened in the past year: restarting after an outage, swapping a unit, pulling firmware, dealing with the utility, handling a repair return, scheduling an inspection, closing out the month’s numbers. Against each one, write who has personally done it, not who could in principle.
Any row with one name is a dependency. Any row where that name is the same across many rows is the thing to work on first. The result is usually uncomfortable and almost always smaller than feared: on most sites, four or five genuinely critical items sit with one person, not forty.
Move it out of the head, cheaply
The reason this does not get done is that it is imagined as a documentation project. It does not have to be.
- Write while doing, not afterwards. The next time the rare task happens, the person doing it narrates and someone else writes. Documentation produced from memory is thinner and wrong in the places that matter, because the hard parts are the ones that feel obvious in hindsight.
- Record the why, not just the steps. A procedure that says what to do decays the moment conditions change. One that says why it is done that way lets the next person adapt it correctly.
- Keep a decisions log. A plain dated list of changes and the reason for each is the single highest-value artifact a small site can keep, and it takes minutes per entry. It is what stops good ideas being re-tried.
- Make the second person do it for real. Shadowing teaches very little. Having someone else perform the task while the expert watches is what reveals which parts were never actually explained.
- Keep a site quirks list. Informal, specific, unglamorous: this position runs hot, this breaker trips on restart, this unit needs the older firmware. It is the single hardest thing to reconstruct after someone leaves.
- Record the relationships. Who to call, at which company, for what, and what was last agreed with them.
The access problem is the urgent half
Knowledge degrades slowly. Access fails immediately, and it is the part that turns a resignation into an outage.
Pool account credentials, wallet configuration, the monitoring and fleet dashboards, firmware licensing, the utility account login, the domain and DNS for anything you run, and physical keys or badge authority. If any of those sit solely with one individual, the operation does not merely lose knowledge when they leave. It loses control of its own systems, sometimes on the same day.
The fix is unglamorous and should not wait for a departure: an inventory of every account and credential, with a named second holder for each, reviewed when anyone joins or leaves. The security argument for individual accounts and the continuity argument point the same way, which is that shared logins are not the answer. Named accounts with a documented recovery path are.
The notice period is not a handover
When someone does resign, the instinct is to use the notice period to write everything down. It rarely works, for a reason worth understanding: the departing person no longer has the thing that generated the knowledge, which is the daily stream of small problems.
Asked to document what they know, people produce an org chart of the obvious and omit the valuable. Not from reluctance. It is genuinely difficult to enumerate your own tacit knowledge on demand, and a last fortnight spent in a room with a text editor tends to produce a tidy document that answers none of the questions that will actually come up.
What works better in that window is narrow and concrete:
- Transfer access first, on day one of the notice. Not the last day. Accounts, credentials, keys, and a verified check that the replacement can actually log in to each one.
- Work the list, not the memory. Take the task inventory described above and walk each single-name row with someone else physically present. Concrete tasks surface concrete knowledge.
- Record the open loops. What is half-finished, what is on order, what is owed, which supplier conversation is mid-flight, what was promised to whom. This is the information with the shortest shelf life and it is almost never captured.
- Agree a question channel afterwards. Most departures are amicable. A short written agreement that they will answer occasional questions for a defined period, paid if appropriate, costs very little and is worth more than the document.
The deeper point is that a notice period is a bad time to discover the dependency. Everything above works far better as routine practice, when the person is not leaving and the knowledge is being produced rather than recalled.
The owner-operator case is the sharpest version
At many smaller operations the person who knows everything is the owner, and that changes the shape of the problem rather than removing it.
There is no notice period and no handover conversation. The triggers are illness, a sale, or simply wanting to take a holiday without the site degrading. The exposure is also broader, because it runs past operations into banking, contracts, insurance and the entity itself.
It also affects what the business is worth. An operation that cannot be run by anyone except its founder is harder to sell and sells for less, because a buyer is acquiring a job rather than an asset. The same documentation that protects continuity is what makes the operation transferable, which is usually the argument that finally gets it done.
If you are hosting, this is your host’s exposure too
For anyone whose machines sit in someone else’s facility, the same concentration exists one level removed, and it affects you without being visible.
Our colocation due-diligence checklist covers what to inspect physically. The staffing version of the same question is reasonable to ask: how many people can perform a restart after an outage, whether procedures exist in writing, and what happens to your fleet if the site manager leaves next month. A facility that has thought about it will answer plainly. One that tells you their manager has been there for years has described the risk rather than addressed it.
Where to start
If this applies to your operation, and on most sites it does, the first move is not a documentation initiative. It is the list: tasks done in the past year, and who has actually done each one. An hour produces it, and it usually makes the next step obvious.
Then fix access, because that fails fastest. Then write down the quirks, because that is the part nobody can reconstruct. The rest can be done gradually, as the work naturally recurs.
The goal is not a site that runs without expertise. It is a site where losing one person is a setback rather than an event. If you are planning a deployment and want to talk through how it would be staffed and documented so it does not depend on a single individual, get in touch.
Explore Rax Mining
- Bitcoin Miner Hosting — Tiered rates from $0.085/kWh, NatGas from $0.055/kWh
- NatGas MDU Units — 1MW modular datacenter containers
- Mining Profitability Calculator — Estimate your mining returns
- Our Facility — Tour our mining infrastructure

