The first two weeks with an agency: a client's field guide

The first two weeks of an engagement predict the next six months with uncomfortable accuracy. Not because the work done in them is proportionally important — it usually isn't — but because every pattern gets set: how fast decisions get made, whether access arrives in hours or fortnights, whether the client's calendar treats the project as a priority or a penance. A project with a crisp fortnight one almost never ends acrimoniously. A project that spends fortnight one chasing DNS credentials and waiting for a kickoff quorum is, statistically, what week fourteen's awkward call is already about.

This is a field guide for the client side of those two weeks — what a good agency will ask for, why, and what you can do to make the whole thing move at the speed you're paying for.

Before the official kickoff: the access checklist

The single biggest hidden variable in project timelines is access latency. Design and engineering work can start on day one; the useful versions of both need logins, data and documentation that routinely take organisations two to four weeks to produce. So a good agency will send the access checklist before the ink is dry. Ours looks like this:

Analytics and data: read access to web analytics, any product analytics, the current CMS, and the last customer research that exists. We're not auditing on day one; we're absorbing.
Technical: repository access (or a repo walkthrough), hosting and DNS registrar details, staging environment, third-party service accounts (email platform, payment sandbox, search).
Brand: current brand guidelines, the actual font and asset files (not the PDF that references them), photography libraries, past design files in whatever state they're in.
People: a named project owner with decision authority and 3–5 hours a week, plus introductions to whoever controls legal/claims review and whoever owns the content we'll be inheriting.

Two things on this list matter more than the rest. The named project owner is non-negotiable — a project with "the team" as the owner has no owner, and every week-one question will spend three days finding a throat. And the brand asset files matter for an unglamorous reason: the fastest possible week-one demo is built on your real materials, and every day spent reconstructing a logo from a screenshot is a day of your budget spent on archaeology.

If any access will take time — security review, procurement portals, an agency of record that controls the DNS — say so at kickoff, not when asked the third time. Blocked items decay; the only defence is an early, dated list of what's pending.

The kickoff meeting itself: ninety minutes, five outcomes

A kickoff that ends with "great energy in the room!" and nothing else has failed. You should leave kickoff with five concrete artefacts, and it's reasonable to hold the agency to them:

The decision map. Who approves what — one name per decision class, said out loud and written down. The mapping session is mildly uncomfortable and saves the project; the mechanics are in our piece on aligning stakeholders without design by committee.
The decision calendar. Not the full project plan — the next three decision dates. When does the founder see direction? When does legal see claims? Dates on a calendar, before anyone leaves.
Success in six months. One paragraph, agreed in the room, of what the engagement has to have achieved to be called a success — in business terms, not deliverable terms. "The site converts trial signups at 4% and the sales team stops apologising for it" beats "new website, phase one."
The demo cadence. A recurring weekly demo slot, defended like a board meeting. If your side can't commit an audience, say so now and fix it now — the cadence is the project's pulse.
The communication contract. One channel, one weekly note format, response-time expectations both directions. Ninety seconds to agree, and it prevents the week-six "I sent that in an email in May" archaeology.

Notice what's absent: a detailed rework of the proposal. Kickoff is for operating agreements, not re-litigating scope. If the scope still isn't right at kickoff, that conversation is legitimate — but it's a different meeting.

Week one: what the agency is actually doing

Clients sometimes worry that week one looks quiet. It isn't; it's just mostly internal. The team is doing the audit pass (your current site, product, analytics, docs), reading everything you sent, setting up environments, and — this is the part that surprises people — building the first rough thing.

A good agency will show you working software in week one or two. Not the real thing — a skeletal version of one core flow, in your brand's rough colours, populated with your real content where it exists. This is deliberate and you should lean into it: a scrappy demo in week one breaks the seal. Every subsequent demo is easier because the first one was ugly and survived. Agencies that spend the first month in research documents are often buying time; be polite, and ask to see the repo and the staging URL.

Your job in week one is responsiveness. Questions will come fast — "which of these three customer phrases is actually how people talk about this?", "is this legacy page load-bearing or historical?", "can support forward us their twenty most common tickets?" These questions have a half-life. Answered in a day, they steer the work; answered in a week, they're guesses by then.

Week two: the first demo and what to do in it

The first real demo is the most consequential meeting of the engagement, and most clients under-prepare for it. Three instructions:

Bring the approver. Whoever signs off on direction should see raw early work with their own eyes. Every layer of retelling — "I showed it to Sarah and she thought…" — adds distortion, and early work distorts badly because it's easy to mistake roughness for wrongness.

React to the problem, not the pixels. Week-two work is for evaluating whether the agency understands the problem and has a point of view. "You've understood the renewal anxiety better than our internal team articulated it" or its opposite is gold. "Could the button be rounder?" is a week-six sentence. There's a whole discipline to this; our companion piece on giving design feedback that speeds work up has the scripts.

Say the uncomfortable thing early. If the direction feels wrong, week two is the cheapest it will ever be to change — roughly free, in fact. Clients who stay polite until week nine pay for their politeness in revision rounds, and the agency pays in something worse than money: the work gets careful instead of good.

The fortnight-one red flags

From either side of the table, these are the early-warning signs worth acting on immediately rather than hoping past:

The project owner gets reappointed to someone "with more availability" who is three levels further from the decision-makers.
Access requests from week one are still open and nobody on your side has been told why.
The weekly demo gets moved twice in the first fortnight — by either side.
You haven't seen anything made: no staging link, no frame, no sketch, only documents.
The agency's questions are all procedural (logo files, style guides) and none are about your customers, your numbers or your fears.

Any one of these is a ten-minute conversation in week two and a six-week ordeal in month three. Raise it with the producer directly; course-correcting in week two is what producers are for.

If you're still pre-kickoff and assembling your side of the table, the brief-writing playbook covers what to bring, and our approach page shows the cadence you'll be slotting into — reading both makes you the best-prepared client in the agency's history, which is a low bar and worth clearing anyway.

Key takeaways

The first two weeks set every pattern the project will live by: decision speed, access latency, meeting hygiene. Treat them as the project in miniature.
Access arrives before kickoff, not after: analytics, technical, brand files, and a named project owner with real authority and real hours.
Leave kickoff with five artefacts: the decision map, three dated decisions, a one-paragraph definition of success, the demo cadence, and the communication contract.
Week-two work looks rough on purpose. Evaluate problem understanding, not polish — the first demo's job is to break the seal, not to be right.
Say the uncomfortable thing in week two, when changing direction costs nothing. Politeness compounds into revision rounds.
Red flags are ten-minute conversations in week two and ordeals in month three. Raise them with the producer, immediately, by name.

FAQ

Our procurement process means access genuinely takes three weeks. Is the project dead? Not dead, but sequence around it. Tell the agency at kickoff exactly what's gated and they'll front-load work that doesn't need it — research synthesis, architecture sketches, content modelling from exports. What kills projects is discovered latency, not declared latency.

We don't have recent customer research. Should week one wait while we commission some? No. Give the agency support tickets, sales call notes, app-store reviews, onboarding emails — the raw material every company has and doesn't call research. A fortnight of reading tickets beats a quarter of waiting for a research vendor, and it's how a week-one demo gets to use words your customers actually said.

Is it a bad sign if the agency pushes back on our kickoff agenda? Usually it's the best sign you could ask for. An agency with a point of view on how the first fortnight should run has run many of them. Reserve worry for the agency that agrees with everything in week one; agreement is cheap there and gets expensive later.

How much of my time do the first two weeks actually need? Budget five hours a week for the project owner in weeks one and two, skewed to fast replies over long meetings: kickoff, one demo, and a daily fifteen minutes clearing questions. The months where clients "don't have time for the project" are precisely the months the project consumes the most of the client's money.