Guide
How to get a new hire useful in three days instead of three weeks
A new person in a small business usually spends their first fortnight in a strange loop. They cannot do the work because they do not know the answers, the answers exist only in the heads of the two busiest people, and asking costs those people ten minutes each time. Everybody is frustrated by Friday, and a meaningful number of new starters quietly decide in that first week that this was a mistake. Three days is achievable, and getting there is mostly a writing exercise rather than a training one.
Week one is when people decide whether to stay
The expensive part of hiring is not the hiring. It is hiring twice.
Somebody who leaves in the first three months costs you the vacancy again, the training again and the disruption again, and the reasons people give for leaving early are remarkably consistent. They were not clear what was expected. Nobody had time for them. They felt in the way. None of those are about the work, and all of them are decided in the first few days.
- The first day sets the expectation for the job. A day spent waiting for an account to be set up tells somebody exactly how organised this place is.
- Interrupting is uncomfortable, so people guess instead. A new starter who has already asked four questions this morning will guess the fifth answer rather than ask again, and the guess reaches a customer.
- Your best people pay the cost. The person a new hire interrupts is by definition the person who knows things, which is the person whose time is worth the most.
- The pattern repeats every single time. The same questions, the same explanations, in the same order, for every new person, forever, because it is never written down.
What a new person actually needs, in order
Onboarding checklists usually mix four different things together, which is why they are long and still miss the important parts. Separated out, they are:
- Access. Accounts, keys, systems, the building. Entirely administrative, entirely predictable, and the most common thing to be unready on day one.
- Answers. The forty things they will be asked or will need to know. This is the largest category and the one nobody writes down.
- Context. Who the customers are, what the business is actually good at, what the unwritten rules are. Cannot be automated and should not be. This is a conversation with a person.
- Reps. Doing the work with a safety net. The only part that genuinely takes time, and the part that gets squeezed out by the other three.
The three days
This assumes the preparation described further down actually exists. Without it, three days is not realistic and nobody should pretend otherwise.
- Before day one: everything is ready. Accounts created, access granted, equipment set up, first week's schedule visible, and a message sent the week before saying where to go, when, who to ask for and what to bring. This step is entirely automatable and it is the one most often missed.
- Day one, morning: the human half. Someone senior spends real time on what the business does, who the customers are and what good work looks like here. Not a slide deck. A conversation, ideally somewhere the work happens.
- Day one, afternoon: shadowing, with permission to say nothing. Watching someone do the actual job, with no expectation of contributing. The explicit permission matters, because otherwise people perform engagement instead of paying attention.
- Day two, morning: the reference pack. Reading the written answers, on their own, with a list of questions to bring back. This is only possible if the pack exists, which is the whole argument of this guide.
- Day two, afternoon: doing the work, watched. Real tasks, real customers where it is safe, with somebody beside them. The goal is not perfection, it is finding out which parts are genuinely hard.
- Day three: doing the work, checked. Working alone with their output reviewed at the end of the day. They will be slow. Slow and independent is the target for day three, not fast.
- End of day three: twenty minutes, honestly. What is still unclear, what was wrong in the pack, what nobody mentioned. Their answers are the most valuable feedback you will ever get on your own documentation, and they are only available once.
The part that makes it possible: write the answers down once
Everything above depends on one thing, and it is not a system. It is that the answers exist in writing.
Most small businesses have never done this, not out of laziness but because the knowledge is invisible to the people who hold it. If you have done a job for nine years, you genuinely cannot see which parts are non-obvious. That is why the only reliable way to build the pack is to capture it from the questions people actually ask.
- Write down every question a new person asks, for a month. Not the answers, just the questions. Anyone who is asked one adds it to a list. This takes seconds and it is the entire discovery phase.
- Add the questions customers ask. They overlap far more than you would expect, and a new starter needs both sets anyway.
- Answer them in writing, plainly, once. Two or three sentences each. The person who knows the answer writes it, not a manager paraphrasing them.
- Put them somewhere searchable. A shared document is fine and is infinitely better than nothing. What matters is that it is one place and it is findable from a phone.
- Have the newest person maintain it. They are the only one who can still see what is missing, and that ability disappears within about a month.
What to automate and what to leave alone
Onboarding is a good example of a process where automating the wrong half makes it worse. A new starter who spends day one clicking through training modules alone has had a worse first day than one who spent it with a person, even though the modules are cheaper.
| The task | Who does it | Notes |
|---|---|---|
| Accounts, access and equipment | The system | Triggered by the acceptance, done before day one, with a check that it actually worked. |
| Paperwork and documents | The system | Collected and chased ahead of the start date rather than on the morning. |
| The first week's schedule | The system | Sent in advance, with names and locations, so nobody arrives not knowing where to stand. |
| Reference answers | Written by people, served by the system | A person writes it once. After that anybody can find it without interrupting anybody. |
| Reminders and check-in prompts | The system | The day three conversation happens because something reminded a manager, not because they remembered. |
| What the business is for | A person | Non-negotiable, and it should be someone senior. Delegating this tells the new starter exactly how much this matters. |
| What good work looks like | A person | Shown, not described. This is the part that cannot be written down usefully. |
| Feedback in week one | A person | Specific, early and in both directions. |
| The unwritten rules | A person | Every business has them. A new person breaks one in week one and needs to be told kindly rather than left to notice the atmosphere. |
What it looks like once it runs itself
The administrative half is a small build and it is usually bolted onto something that already exists, because it is the same shape as any intake process.
An offer is accepted. That single event creates the accounts, grants the access, orders the equipment, sends the welcome message with the practical details, collects the documents and chases them, puts the first week in a calendar, and tells the manager what is outstanding three days before the start date.
On day one, nothing is missing. On day two, the questions have written answers that are searchable from a phone in a van. On day three, a prompt reminds somebody to have the conversation that most businesses skip.
None of that is clever. It is the same argument as everywhere else on this site: the parts of a process that repeat identically every time should not depend on a busy person remembering them, and the parts that require judgement should never be handed to a system.