A new agency hire should ship something real to a client in their first week and own a small client relationship within the first month. The three things that determine whether onboarding works are having actual work ready on day one, naming a specific person as their point of contact for questions, and setting the delegation boundary explicitly - what they decide alone, what they check, what always escalates. Most agency onboarding fails not through neglect but through vagueness: the new person is welcomed warmly, given access to everything, and left to work out what they are supposed to do.
The first month sets how someone works for the next three years.
Not because of what they learn in it - most of that is forgotten - but because of what it teaches them about how things are done here. Whether questions are welcome. Whether it is acceptable to say something is unclear. Whether the standard is genuinely the standard or an aspiration.
Agencies are usually good at the welcome and poor at the structure. The new person is greeted warmly, added to every channel, told to shout if they need anything, and then left in a fog for three weeks that nobody notices because everybody is busy.
This guide covers the week before they start, the first day, the first week, the first month, the delegation conversation, and the 30-60-90 reviews.
Before they start
Four things, none of which takes long and all of which are usually left to day one.
Have work ready. The single most important item. The most common onboarding failure is hiring for future capacity and having nothing concrete on Monday. Line up their first real task before they accept, not after they arrive.
Set up access in advance. Email, systems, tools, calendar. A new person spending their first afternoon waiting for logins learns that things here are disorganised, and it is a needlessly bad first impression.
Name their point of contact. One person, explicitly, whose job for the first month is answering their questions. Without a name, a new hire distributes questions thinly to avoid bothering anyone, and gets slow answers from everyone.
Tell the team who is joining and what they will own. Obvious, frequently missed, and it prevents the awkward first week where nobody is sure what the new person does.
Day one
Keep it short and concrete. A full day of orientation is exhausting and most of it does not stick.
Morning: people and context. Who does what, who to ask about what. The clients, briefly - who they are and what you do for them. Not a deep dive; they cannot absorb it yet.
Afternoon: something real. Give them a small, genuine task on a live project. Something low-risk that ships.
That afternoon is deliberate. Doing something real on day one does more for confidence and belonging than any amount of orientation, and it immediately surfaces the practical gaps - a missing permission, a tool nobody mentioned - while there is time to fix them.
End the day with fifteen minutes. What was confusing? That question, asked on day one and answered honestly, is worth more than a handbook.
Week one
The goal is one thing: ship something to a client.
Small, reviewed, but genuinely delivered. It tells them the work is real and tells you how they work.
Alongside that:
Walk them through one project end to end. Not the work - the process. How it was sold, scoped, kicked off, delivered, invoiced. Seeing one complete cycle explains more than any process document, and it is the fastest way to convey how your agency actually operates.
Show them the commercial shape. How you price, what a good margin looks like, what absorbing out-of-scope work costs. Agencies routinely keep this from delivery staff and then wonder why people agree to extra work without flagging it. Someone who does not know the project is fixed-fee has no reason to treat scope as finite.
Introduce them to clients they will work with. Briefly, on an existing call. Being a name and a face early makes the first real interaction much easier.
Book the 30-day review now, so it is in the calendar rather than dependent on someone remembering.
The delegation conversation
The single highest-value conversation in onboarding, and most agencies never have it explicitly.
Write down three categories:
What they decide alone. Day-to-day craft choices, small client questions within scope, how they organise their own work.
What they check first. Anything affecting timeline, anything a client might perceive as a change, anything they are unsure about. Name who they check with.
What always escalates. Scope changes, pricing, complaints, anything involving money, anything where a client is unhappy.
Without this, one of two things happens. They check everything, which recreates the bottleneck you hired to remove and makes them feel untrusted. Or they decide things they should not have, and someone finds out afterwards - which damages confidence on both sides for something that was never their fault.
Ten minutes, written down, revisited at 30 days as the boundary widens.
Month one
Give them a small client relationship to own. Low-risk, small, genuinely theirs - they run the updates, answer the questions, own the outcome.
Ownership accelerates competence far faster than shadowing. Someone who has owned one small thing for three weeks understands the job better than someone who has watched five people do it.
Have them run something end to end. A small project or a discrete phase, from brief to delivery.
Get them into the rituals properly. Standups, kickoffs, retrospectives - not as an observer but with a part to play. Retrospectives are particularly valuable early: a new person sees process problems that everyone else has stopped noticing, and that window closes within about two months.
Ask what is still unclear, repeatedly. New people underreport confusion, especially in weeks two and three when they feel they should have worked it out. Ask specifically: "what is the thing you have been quietly unsure about?"
The 30-60-90 reviews
Short, scheduled, both directions.
30 days. Mostly about them. Is the role what they expected? What is unclear? What do they need? Light on performance - it is too early - and heavy on removing obstacles. Widen the delegation boundary here if it is warranted.
60 days. Balanced. Specific feedback on work, both directions. This is where you address anything that is not working, while it is still easy to correct and before habits set.
90 days. The honest conversation. Is this working, for both parties? By ninety days you know, and so do they. If it is not working, saying so now is far kinder than waiting for a formal review in month six.
Keep them to thirty minutes and keep them scheduled. Reviews that depend on someone feeling they are needed do not happen.
What to write down, and what not to
Worth documenting: how a project runs from sale to invoice, your QA checklist, the client list with a paragraph each, tool access and conventions, the delegation boundary, and who to ask about what.
Not worth documenting: everything else. A comprehensive handbook is a large investment that goes stale and that nobody reads twice. A short, current document beats an exhaustive, outdated one.
The most useful artefact is usually a one-page "how we work here" covering the non-obvious conventions: how requests arrive, what "done" means, when to raise something, how feedback is given. Those are the things people are most uncertain about and least likely to ask.
Have the newest person update it. They can see what was missing; in three months they will not be able to. This is the cheapest way to keep onboarding documentation current, and it improves with every hire.
Remote and distributed
Two adjustments.
Over-invest in the informal. Most of what a new person learns in an office is ambient - overheard conversations, watching how someone handles a difficult client. Remotely that is absent, and it must be replaced deliberately: recorded client calls to listen to, invitations to conversations they could technically skip, a scheduled informal chat each week with someone different.
Make asking easy and asynchronous. A named contact matters more remotely, because the barrier to interrupting is higher. Say explicitly that questions are expected, and answer them quickly enough that the message is credible.
The first-week schedule, concretely
A workable structure, because "onboard them well" is not actionable and a timetable is.
Monday. Morning with their point of contact - people, clients, how work flows. Afternoon on a small real task, with the explicit instruction to note down everything confusing. Fifteen minutes at the end to go through that list.
Tuesday. Walk one completed project end to end - how it was sold, scoped, kicked off, delivered, invoiced. Then their first piece of work on a live project, reviewed by a peer before it goes anywhere.
Wednesday. The commercial briefing - how you price, what a good margin looks like, what absorbing out-of-scope work costs. Then the delegation conversation, written down. Introduce them on one client call as an observer.
Thursday. Independent work with their reviewer available. This is the first day they should feel slightly unsupervised, deliberately.
Friday. Ship something to a client. Then thirty minutes: what was confusing, what is still unclear, what they need next week. Book the 30-day review before they leave.
That is five days that produce a person who has delivered something real, seen a full project cycle, understands the commercial model, and knows what they are allowed to decide. Compared with the common alternative - access granted, handbook sent, three weeks of fog - it costs perhaps four hours of other people's time.
Onboarding into an existing client relationship
A specific case worth handling deliberately, because it is where new hires are most likely to stumble through no fault of their own.
Brief them on the history, not just the current state. What has gone wrong before, what the client is sensitive about, which decisions were contested. A new person who cheerfully reopens a settled argument in their first client call has been set up to fail.
Introduce them properly and explain the change. Clients dislike discovering a new face with no explanation, and they read it as instability. A short message beforehand - who is joining, what they will own, why it is good for the client - prevents it entirely.
Do not hand over the relationship immediately. Overlap for at least a few weeks, with the existing owner present. Relationships transfer through shared context, not through an introduction email.
Onboarding someone senior
A different problem from onboarding a junior, and applying the same process to both is why senior hires frequently take longer to become effective than they should.
Give them a real problem, not a warm-up. A senior person given trivial work for three weeks concludes they were hired for a role that does not exist. Hand them something genuinely unresolved in week one - a client relationship that needs attention, a process that is not working - and let them form a view.
Ask what they see, early and specifically. A senior hire's most valuable output in the first month is their observations, and that window closes within about eight weeks as they normalise to how things are done. "What looks wrong to you?" asked at week three gets answers you will not get at week twelve.
Be explicit about what is settled and what is open. The most common senior-hire failure is proposing changes to things that were decided deliberately for reasons nobody explained. Saying "pricing and positioning are settled, delivery process is genuinely open" saves months of misdirected energy.
Do not make them prove themselves before granting authority. A senior person operating without decision rights is expensive and visibly awkward to the team, who cannot tell whether to treat them as an authority. Grant the authority the role requires from the start and manage the risk with the 30-day review rather than by withholding.
When onboarding reveals a hiring mistake
Occasionally the first month makes clear the fit is wrong. Handling that well matters, and delay is the enemy.
Distinguish an onboarding gap from a hiring mistake. Someone struggling because nobody briefed them properly is not a bad hire, and the honest test is whether the things they are getting wrong were ever explained. Most early struggles are the former.
Raise it at 30 days, not 90. Specific, factual, with the support you are offering. Someone told at thirty days that a particular thing is not working has a genuine chance to fix it; someone told at ninety has spent two months being quietly assessed.
If it is not working at 90 days, say so plainly. Continuing past that point in the hope of improvement is unkind to them - they are in a role they are not succeeding in, and they can feel it - and expensive for you. A clear, respectful conversation at ninety days is far better than a drawn-out six months.
What good looks like at 90 days
A useful test rather than a feeling. At ninety days a well-onboarded agency hire should be able to:
- Run a small project end to end without checking each step
- Answer a routine client question without escalating
- Explain why the agency prices the way it does
- Recognise an out-of-scope request and know what to do with it
- Name who to ask about anything they cannot answer
If several of those are missing at ninety days, it is far more likely to be an onboarding gap than a hiring mistake - and it is still cheap to fix at that point, which is exactly why the ninety-day conversation matters.
Building the onboarding you wish you had had
The cheapest way to build a good onboarding process is to have every new hire build the next one.
Ask them to keep a confusion log. From day one, a running note of everything unclear, every unwritten convention they had to discover, every question they hesitated to ask. It costs them nothing and it is the single most valuable onboarding artefact you can produce, because they can see what is missing and in three months they will not be able to.
Review it at 30 days and turn it into documentation. Most entries become a line in the "how we work here" page. Some reveal genuine process gaps that affect everyone, not just new people.
Have them own the next person's first week. Someone three months in remembers exactly what was hard, and giving them the responsibility makes it concrete. It also develops them, which is a useful side effect at no cost.
This approach means the process improves with every hire rather than being written once and going stale, and it removes the need for anyone to sit down and author a handbook in the abstract - which is a task that never rises to the top of anyone's list.
What onboarding costs, and what it saves
Worth quantifying, because onboarding competes with billable work and loses unless someone has done the arithmetic.
The cost. Roughly four to six hours of other people's time in week one, plus perhaps two hours a week for the following month. Call it fifteen hours, mostly from one or two people.
The saving. The difference between someone productive at six weeks and someone tentative at sixteen. On a delivery hire generating around 1,000 billable hours a year, ten weeks of reduced effectiveness is a substantial number - and that is before counting the cost of a hire who leaves at four months because they never found their footing.
The risk it removes. Early attrition is the expensive failure. Recruitment cost, lost ramp time, and the team absorbing the gap again. Structured onboarding does not prevent every bad fit, and it does prevent the ones caused by someone concluding, reasonably, that nobody here knew what they wanted from them.
Fifteen hours against that is not a close decision, and it is worth saying out loud to whoever is being asked to give up the time.
The mistakes
No work ready. The most common and the most damaging to confidence.
Access not arranged. Small, avoidable, and it sets a tone.
Nobody named. Questions distributed thinly and answered slowly.
Correcting everything. If you correct every deviation from how you would have done it, you train them to check everything, which rebuilds the bottleneck. Correct what matters; let the rest be different.
Assuming they will absorb the commercial context. They will not, unless you tell them, and it directly affects how they handle scope.
No 90-day conversation. Problems that were addressable at ninety days become entrenched by month six.
Remote-specific practices worth adopting
Beyond the two adjustments already covered, three practices consistently help distributed hires.
Record client calls and give them a back catalogue. Three or four recordings of real client conversations teach tone, expectations and vocabulary faster than any briefing document, and they can be watched at their own pace.
Pair them on something in the first fortnight. Two people on one task for ninety minutes, screen shared. It transfers the unwritten conventions - how carefully things get checked, what "good enough" means here - that never make it into documentation.
Schedule the informal deliberately. A weekly half-hour with a different person each time, with no agenda. In an office this happens by accident; remotely it does not happen at all unless someone puts it in a calendar.
The summary
Onboarding is not a document, it is a first month designed on purpose.
Have work ready. Name one person to answer questions. Ship something to a client in week one. Set the delegation boundary explicitly. Give them a small relationship to own by week four. Review at 30, 60 and 90 days.
None of it is elaborate, and it is the difference between someone productive in six weeks and someone still tentative in six months - which, at the cost of a hire, is a substantial amount of money for something that takes a few hours to set up.