Client Kickoff to Delivery: Agency Project Workflow
The projects that go wrong rarely go wrong in week three. They go wrong in the kickoff call, or in the four days after it, and nobody notices until the deadline everyone agreed to stops looking realistic. I have spent twelve years running delivery operations for marketing agencies across 92 completed engagements, and across all of them the same pattern holds: the first week decides more than any stage that comes after it.
I have written up the seven-stage process a project moves through from intake to reporting in the agency delivery process, and I am not repeating that here. This is the part before the stages start counting: what actually happens in the kickoff room, what gets built inside the tool that same day, the checkpoints across the first week, and the difference between a project that has started on a calendar and a project that has actually started moving.
What has to happen in the kickoff call itself?
A kickoff call has one job: turn what was sold into something a team can start without guessing. Most kickoffs fail at that job while feeling productive, because the conversation drifts toward the relationship rather than the work.
The agenda that actually prevents rework is short and specific.
- Definition of done, stated out loud. Not the proposal language, the actual deliverable, its format, and how anyone in the room would judge it finished.
- What is explicitly excluded. Said in the same breath as what is included, so it is a decision rather than a discovery three weeks later.
- One named owner on each side. Not a department, not “the team,” a person who answers when something stalls.
- The approval chain. Who signs off on the client side, and who consolidates feedback into one voice before it reaches you.
- The first real date. Not the final deadline, the date something specific is due to move, because that is the date that tests whether the project is actually underway.
Notice what is missing from that list: a walkthrough of your process, your methodology, or the tool you use internally. None of that belongs in the client’s hour. It answers a question they did not ask and uses time that should go to the five items above.
Who actually needs to be in the room?
The person who will do the work needs to hear the kickoff directly, not receive a summary of it afterward. I have watched more rework get created by a clean handoff from a salesperson to a producer than by any client behaving unreasonably. The producer inherits a paraphrase of what was said, fills the gaps with a reasonable guess, and the guess is wrong often enough to matter.
On the client side, the person who consolidates feedback needs to be named in the call, not assumed. If that is unclear going in, ask directly. An agency chasing three separate opinions through week two is paying for a question that should have been asked in week one.
What gets built inside the tool before the call ends?
A kickoff that produces a good conversation and nothing in the system has produced nothing. The test I use is simple: could someone who was not on the call open the tool tomorrow and know what to do. If the answer is no, the call was a conversation, not a start.
What should exist by the time everyone logs off:
- A project record, created live during the call, not typed up afterward from notes.
- The owner field filled with a name, not a role.
- A brief started inside the project record itself, holding the definition of done and the exclusions from the agenda above.
- An intake form sent the same day, asking for access and assets rather than information you already have.
- A first task, dated, assigned, and visible on whoever is doing the work’s board before they log off for the day.
These are the same artifacts I have written up in detail, field by field, in the templates that make a delivery process repeatable. The point here is narrower: they get created on kickoff day, inside the live project record, not assembled later from memory of what was discussed.
The first-week checkpoints
Between the kickoff and the end of week one, I run the same rhythm on almost every engagement, adjusted for the size of the project.
Day one
The kickoff happens, the project record exists with an owner and a first task, and the intake form goes out before the day ends. If the intake form is still sitting in a drafts folder the next morning, the project has not actually started, whatever the calendar says.
Day two to three
This is the access-and-assets window, and it is where projects quietly stall without anyone flagging it. Logins do not arrive, brand files are half sent, the approval contact goes quiet. The discipline is not chasing harder, it is checking the intake form against what is still blank and naming the gap to the client directly rather than working around it.
Day five
A status update goes out at the end of the first week regardless of how much visible progress there was. This is the one habit that separates agencies clients trust from agencies clients worry about. A quiet first week with no update reads as nothing happening. A quiet first week reported honestly, blocked on assets, next step named, reads as a project under control.
What separates a project that has started from one that has only been scheduled
A project can look started and not be. The call happened, a folder exists, a due date sits in the tool. None of that is the same as underway.
The test I actually use: has anything moved inside the tool since the kickoff that was not moved by me chasing it. A task assigned and picked up on its own, an intake field filled without a reminder, a status changed by the person doing the work rather than the person managing it. That is the project running under its own motion instead of mine. Until something moves without a push, the project is nominally started and nothing more, and the schedule everyone agreed to is still theoretical.
Where week one goes wrong even with a good agenda
A few patterns account for most of the week-one failures I have seen, and they are worth naming plainly because they are avoidable.
- The kickoff gets booked before the assets exist. A date on a calendar is not readiness. If brand files and access are not confirmed, the call still happens, but it becomes an assets meeting wearing a kickoff’s name, and the real kickoff has to happen again once they arrive.
- No client-side owner gets named. The call ends warm and everyone assumes someone will follow up. Nobody owns following up, so nobody does, and the project drifts for a week before anyone admits it.
- The first task sits with the salesperson instead of the producer. The handoff from sold to started happens by memory rather than by a task assigned in the tool, and detail gets lost in the retelling every time.
Each of those is a choice made in the room, not a client behaving badly. That is the useful part: they are fixable by changing what happens in the first hour, not by hiring differently or working harder in week three.
If you want this run for you rather than run by you
Some agencies want the choreography above handled as a standing function rather than owned internally, particularly once there is more than one kickoff happening in the same week. That coordination, intake, scheduling, and the first-week checkpoints, run as a defined service rather than a role you hire for, is what I describe on the page for project management delivered as a service.
Whichever way you run it, the diagnosis usually comes first. The Agency Ops Audit is a $1,500 fixed-fee review, two weeks, ending in a written 90-day roadmap and a 60-minute readout, that looks at how your projects actually start today and names where week one is losing you time before anything gets rebuilt. If the honest finding is that your kickoff already works, that is what the readout will say.
