Client onboarding should be boring, repeatable, and mostly finished before kickoff

Every agency has an onboarding process. Most of them have it in one person’s head, which means it exists right up until that person is on holiday or busy or has simply had a long week.

The cost of improvised onboarding is not the improvising. It is that the things you forget are always the same things, and they are always the ones that hurt later: access, decision rights, and the definition of done.

What onboarding is actually for

It is tempting to think of onboarding as hospitality — welcome pack, kickoff call, nice deck. Those are fine. But the operational purpose is narrower and more useful. Onboarding exists to answer four questions before work starts:

• Who decides? Which named human can approve scope and sign off on work. Not “the client.” A person.
• What did we agree? An estimate with a number, a date range, and an approval on the record.
• What do we need from them? Access, assets, credentials, content — listed, owned, and dated.
• What does done look like? For the first milestone at minimum.

Every miserable project you have ever run failed at least one of those in the first fortnight.

A checklist that survives contact

Here is a version that works for most small studios. Adapt the specifics; keep the shape.

Before kickoff

1. Client record created — company, contacts, billing contact if different.
2. Project created and typed: one-time, retainer, maintenance, or block of hours. The type changes how you will run it.
3. Estimate drafted, submitted, and approved — with start and end dates. If it is not approved, kickoff does not happen.
4. Named approver confirmed in writing, including who covers when they are away.
5. Access request sent as a dated list with an owner per line.

Kickoff week

1. Milestones created with dates, not just a deadline at the end.
2. Team assigned, with each contractor able to see the estimate they are working against.
3. Communication rhythm agreed: where updates land, how often, and what an emergency means.
4. First review date booked in the calendar before anyone needs it.
5. Kickoff time logged like real work — because it is, and because it is where margin quietly goes.

End of week two

1. Access fully working. Not “mostly.” Chase anything outstanding now, while goodwill is high.
2. Burn checked against the estimate for the first time.
3. Any favours already granted turned into visible tasks.

If a step on your checklist has been skipped on three consecutive projects and nothing bad happened, delete it. A checklist people ignore is worse than no checklist, because it teaches your team that the process is optional.

Make the system carry the process

Written processes decay because they live in a document nobody opens. The ones that last are the ones the tool enforces by structure.

Concretely: if a project cannot move past “pending approval” until an estimate is approved, you do not need a checklist item for it. If every estimate has start and end dates as fields rather than as prose, the scheduling conversation happens automatically. If tasks require an assignee, the “who owns this” question resolves itself.

The rule of thumb: anything you can turn into a field or a state, turn into a field or a state. Reserve the written checklist for the genuinely human parts — the conversations, the expectation-setting, the awkward question about who signs off.

The awkward questions worth asking early

These feel uncomfortable in week one and save weeks in month three:

• “Who else needs to approve this before it is final?”
• “If we are waiting on feedback, how long should we wait before we chase, and who do we chase?”
• “What happens to the timeline if the content arrives two weeks late?”
• “Is this budget approved, or is it the number you are hoping for?”

Clients almost never mind being asked. They mind being surprised.

Common questions

How long should onboarding take?

Two weeks from signature to real work is normal, and most of it is waiting on access. Compress it by sending the access list before the contract is signed, not after.

Do small projects need all this?

Scale it, but do not skip the approved estimate and the named approver. Those two are cheap on a small project and catastrophic to be missing on any project.

What if the client will not do their part?

Record it as a delivery phase — waiting on client — with a date, and report it back neutrally at the end. A client who sees they held the project for eleven days behaves differently on the next one.

The goal is not a thicker process. It is that the second project with a client feels identical to the first, no matter who is running it.