A kickoff meeting exists to agree how decisions get made, not to present a plan. The seven items on the agenda are: why this project exists, what success looks like in numbers, who approves what, the timeline with the client's own dependencies named, how you will communicate, what is explicitly out of scope, and what happens when something changes. Run it for 60 to 90 minutes with the approver in the room, and send a written summary within 24 hours. The most valuable output is not the plan - it is agreement on the approval path, because almost every late project can trace part of its delay to waiting on a decision from someone who was never named.
Most kickoff meetings are presentations. The agency walks through a timeline, the client nods, everyone feels positive, and the project starts.
Six weeks later the same project is stuck because feedback arrived from three people who disagreed, the content nobody was assigned never appeared, and the person who can actually approve the design has been on leave for a fortnight.
None of that was unforeseeable. All of it was decided - by omission - at a kickoff that covered what would be built and not how the two organisations would work together.
This guide covers what a kickoff is actually for, the seven-item agenda, who needs to be in the room, the questions that surface problems early, and what to send afterwards.
The one-sentence version
A kickoff is where you agree how decisions get made. Everything else on the agenda supports that.
What a kickoff is for
The reframe that changes everything: a kickoff is about the working relationship, not the work.
The work is already defined. It was scoped in the proposal, priced, and agreed. Re-presenting it is comfortable and adds nothing.
What has not been agreed is the machinery: who decides, how fast, through what channel, and what happens when something moves. That machinery determines whether the project runs smoothly far more than the plan does, and it is the only thing you cannot fix later without a difficult conversation.
Three things a good kickoff produces:
A named decision-maker, with a named deputy. A shared understanding of what the client has to do, with dates. An agreed process for change, established before anything needs changing.
Everything else is useful context.
The ninety minutes that decide the next three months
A kickoff costs a small amount of a few people's time. What it prevents - rework from contradictory feedback, a fortnight lost to an absent approver, a scope dispute at delivery, a case study with no baseline - each cost considerably more.
And unlike most process improvements, it needs no tooling, no budget and no change to how anyone works. Just an agenda and the willingness to read the exclusions out loud while everyone is still pleased with each other.
Who should be there
The approver. Non-negotiable. A kickoff without the person who signs things off is a kickoff whose main output cannot be produced. If they cannot attend, move it.
Your delivery lead, not only the account lead. The person doing the work should hear the objectives first-hand rather than through a relay, and clients notice when the people who sold the project are the last ones they ever see.
Anyone with a dependency. If the client's developer, IT team or legal function will need to do something, they belong in the room now rather than being discovered in week five.
Keep it under eight people. Beyond that it becomes a presentation again, because nobody speaks freely in a large group.
Why the meeting is worth protecting
Kickoffs get compressed, postponed and occasionally skipped, usually because the project is already behind before it starts and ninety minutes feels indulgent.
That instinct is exactly backwards. A project starting late is a project with less slack, which makes the things a kickoff establishes - a named approver, agreed turnaround times, explicit exclusions - more valuable rather than less. The projects that most need a proper kickoff are precisely the ones under pressure to skip it.
The seven-item agenda
Sixty to ninety minutes. Send it in advance.
1. Why this project exists (10 min)
Not the scope - the reason. Ask the client to describe it in their own words, even though you already know.
Two things come out of this. You frequently learn the real driver, which differs from the stated one - a board commitment, a competitor, a personal reputation on the line. And the wider group hears it, which matters when someone three weeks in questions a decision.
The question that works: "If this goes perfectly, what changes for you?"
2. What success looks like, in numbers (10 min)
Agree the measurable outcome and write down the current baseline.
This is the most commonly skipped step and it has an outsized consequence: without a baseline captured at the start, you will not be able to demonstrate the result at the end. Ten minutes now is the difference between a case study with numbers and one without, as covered in our guide to testimonials and case studies.
If there is no number, agree a qualitative test that both sides would recognise. "The sales team sends prospects to the site without caveating it" is a legitimate success condition.
3. Who approves what (10 min)
The highest-value item on the agenda.
Establish:
- One named approver for each type of decision. Design, copy, technical, commercial - possibly different people.
- A deputy for each, for leave and illness. Two weeks of absence with no delegate has stalled more projects than any technical problem.
- How feedback is consolidated. Anyone can give feedback; one person reconciles it. Contradictory feedback from unnamed stakeholders is one of the largest hidden costs in agency delivery.
- How long approvals take. Agree a turnaround, in working days.
Ask directly: "If we send something for approval on a Monday, who looks at it and when will we hear back?" The answer, and the hesitation before it, tells you a great deal.
4. The timeline, with their dependencies named (15 min)
Walk the plan, but spend the time on what the client owes and when.
Content, access, credentials, sample data, brand assets, legal review, third-party coordination. Each one with a name and a date.
Then say the arithmetic out loud, once, neutrally: "Each week these slip, the delivery date moves by a week, because the team scheduled for the next phase moves on."
Saying it at kickoff, when nobody is defensive, is enormously easier than raising it in week six when the date is already at risk. It also converts lateness from a complaint into a shared timeline, which is what it actually is.
5. How we will communicate (10 min)
Cadence, channels, response times. This is the communication plan, agreed here rather than assumed.
The one rule worth stating explicitly: if a request is not in the project system, it is not a request. Say it kindly and early, because you will be enforcing it for the next three months.
6. What is out of scope (10 min)
Read the exclusions aloud. All of them.
This feels awkward and it is the cheapest insurance available. An exclusion buried in a proposal appendix that nobody read is not an agreement; the same exclusion said out loud, with the approver present, is.
It also surfaces mismatches while they are free to fix. If the client visibly assumed something excluded was included, you want to discover that now - either you add it as a priced change or you correct the expectation, and both are far better than the discovery happening at delivery.
7. What happens when something changes (5 min)
Explain the change process in two sentences. Anything outside scope gets a quick estimate before work starts, so they can decide each time.
Framing matters. This is not a defensive mechanism, it is how they retain control of the budget: "You'll never get a surprise invoice, because nothing outside the scope happens without you approving it first." True, and much better received than a clause read out.
Our guide to change orders covers making the process light enough that using it is easier than absorbing the work.
The two things to get right if you get nothing else
A named approver, with a named deputy. Almost every late project can trace part of its delay to waiting on a decision from someone who was never identified, or who was on leave with no delegate.
The exclusions, read aloud. An exclusion buried in an appendix nobody read is not an agreement. The same exclusion said out loud, with the approver present, is - and it surfaces mismatches while they are still free to fix.
Everything else on the agenda is valuable. Those two are the ones that pay for the meeting on their own.
Four questions that surface problems early
Beyond the agenda, four questions consistently reveal risk while it is still cheap.
"Has anyone tried to solve this before?" A failed previous attempt is enormously informative - about the organisation, the constraints, and what will be politically difficult.
"Who in your organisation is sceptical about this?" There is almost always someone. Knowing who, and why, is far better than discovering them in a review three weeks in. Ask it lightly.
"What would make you cancel this project?" Uncomfortable and clarifying. The answer tells you what to protect.
"What has annoyed you about working with agencies before?" People answer this honestly and specifically. It is the fastest route to understanding their expectations, and it costs nothing to ask.
Timing it
Run the kickoff after the contract is signed and before any work starts. Both halves matter: before signature it is a sales meeting and people commit to nothing; after work has begun the decisions it exists to make have already been made by default.
The internal kickoff
Run one before the client kickoff. Thirty minutes with the delivery team.
Cover the scope and its exclusions, the commercial shape - fixed fee, retainer, budget - so the team understands what absorbing extra work costs, the risks you can already see, and who owns what.
Two things this prevents. Team members agreeing to out-of-scope requests because they had no idea the project was fixed-fee. And the delivery team learning the objectives second-hand, filtered through whoever sold it.
The document is the agreement
Everything decided verbally in a kickoff decays at the same rate as everything else people agree verbally. The written summary is what converts ninety minutes of goodwill into something you can point at in week six, when a detail is contested and both sides genuinely remember it differently.
Send it within 24 hours, invite correction, and treat silence as agreement. That last part is why the invitation matters - it makes the absence of a reply meaningful rather than ambiguous.
What to send afterwards
A written summary within 24 hours. Not minutes - decisions.
Kickoff summary - Northwind website rebuild
Objective. Increase demo request conversion from 1.2% to 2.5% by end of Q1. Baseline recorded 14 Aug.
Approvals. Sarah Chen approves design and copy; Mark Ellis deputises. Technical decisions with Dev Patel. Turnaround agreed at 3 working days.
From Northwind. Product copy by 3 Sep (Sarah). CMS credentials by 22 Aug (Dev). Brand assets in editable format by 22 Aug (Sarah).
From us. Wireframes 5 Sep, designs 26 Sep, build complete 7 Nov, launch 14 Nov.
Communication. Written update every Wednesday. Fortnightly call, Tuesdays 10am. Requests in the project workspace.
Out of scope. Blog content, historical post migration, integrations beyond the CRM, ongoing hosting.
Changes. Anything outside scope gets an estimate before work starts.
Please flag anything that looks wrong by Friday.
That document is the reference point for the next three months. When something drifts - and it will - you are pointing at an agreement rather than a recollection.
The closing line matters. Inviting correction converts silence into tacit agreement, which is what you need it to be.
Book it before the contract is signed
Put the kickoff in the diary while the client is still enthusiastic about starting. Arranging it afterwards competes with everyone's calendar and slips by a fortnight.
Preparing for it
An hour of preparation makes the difference between a kickoff that agrees things and one that discusses them.
Send the agenda 48 hours ahead. Including the exclusions list. Some clients will read it and raise a mismatch by email, which is a better place to discover it than in the meeting.
Pre-fill what you can. Bring a draft timeline, a draft list of client dependencies with proposed dates, and a proposed approver. People edit a draft far more readily than they generate one from nothing, and it turns a 90-minute discussion into a 60-minute confirmation.
Do the internal kickoff first. Thirty minutes with the delivery team, covering scope, exclusions, commercial shape and the risks you can already see. Walking into the client meeting with your own team aligned is what lets you commit to things in the room.
Know your two hardest questions. There are usually one or two things you are unsure about - a dependency, a date, a stakeholder. Decide in advance how you will raise them, because the temptation in a positive meeting is to let them slide.
What to do if it goes badly
Occasionally a kickoff surfaces something serious - a scope mismatch, an unavailable approver, a stakeholder who was not consulted before the project was bought.
Do not paper over it to keep the mood. A kickoff that ends warmly with an unresolved contradiction is worse than one that ends awkwardly with it named, because the contradiction does not go away and it costs more later.
Stop and re-scope if the gap is material. A project that starts on a misunderstanding runs on one. Saying "I think we should pause and align on this before we start building" is uncomfortable for ten minutes and saves weeks.
Treat an absent approver as a blocker. Rescheduling the kickoff is a small cost; running a project with no named decision-maker is the largest single predictor of delay there is.
Adapting it to the engagement
The seven items hold. The weighting changes.
Small projects (under four weeks). Compress to 45 minutes and cut the timeline walk-through, but keep items 3, 6 and 7 - approver, exclusions, change process. Those three are where small projects go wrong, and their brevity is exactly why agencies skip the kickoff entirely and then absorb a fortnight of unscoped work.
Retainers. The agenda shifts substantially. There is no delivery date, so the timeline item becomes how work gets requested, prioritised and queued. Add the allocation, the rollover rule and the overage policy explicitly, since those are the terms most likely to be misremembered later. See retainer models.
Repeat clients. Tempting to skip, and worth a shortened version anyway. People change roles, approval structures shift, and the assumptions carried over from the last project are frequently the ones that cause problems. Twenty minutes confirming the approver and the exclusions is enough.
Long or phased programmes. Run a full kickoff for the programme and a short one at each phase. The phase kickoff is mostly item 3 and item 4 - who approves this phase, and what do we need from you by when.
Reading the room
Beyond the agenda, a kickoff is the first real opportunity to observe how the client organisation actually operates, and three signals are worth noting privately.
How they handle disagreement. If two people from the client's side disagree in the meeting, watch how it resolves. That resolution mechanism is the one your project will run on for the next three months.
Whether the approver defers. An approver who repeatedly says "we'll need to check with someone" is not the approver, whatever the org chart says. Find out who is, before you build a process around the wrong person.
How specific they are about the objective. A client who cannot say what success looks like beyond generalities is not being evasive; they usually genuinely have not decided. That is a project risk worth naming early, and frequently a paid discovery phase is the right answer.
None of these need raising in the meeting. All three should change how you plan the engagement.
Send the agenda first
Forty-eight hours ahead, including the exclusions list. Some clients will read it and raise a mismatch by email, which is a considerably better place to discover one than in the room.
The mistakes
No approver present. Covered above, and the single most common one.
Presenting instead of agreeing. If you spoke for fifty of sixty minutes, it was not a kickoff.
Skipping exclusions because the mood is good. The mood is good precisely because nothing has gone wrong yet. That is the moment to have the conversation, not a reason to avoid it.
No dates on client dependencies. "You'll send us the content" is not a commitment. "Sarah sends product copy by 3 September" is.
Not sending the summary. Everything agreed verbally decays at roughly the rate of everything else people agree verbally.
Why this is the cheapest hour in the project
A kickoff costs ninety minutes of a few people's time. The things it prevents - rework from contradictory feedback, a fortnight lost to an absent approver, a scope dispute at delivery, a case study with no baseline - each cost considerably more than that.
And unlike most process improvements, it requires no tooling, no buy-in beyond the room, and no change to how anyone works. It requires an agenda and the willingness to spend ten minutes reading the exclusions out loud while everyone is still pleased with each other.