The weekly operating rhythm that prevents client surprises

Almost no client relationship goes wrong suddenly. The angry email in week ten is the visible end of something that started in week four, when a deadline moved by three days and nobody flagged it, and then again in week six, when an approval sat unanswered and the team quietly worked around it.

The fix isn't more meetings. It's one meeting, at the same time every week, that asks the same short list of questions and produces decisions. Agencies that run this find problems roughly a fortnight earlier than agencies that don't, which is usually the difference between a change request and a discount.

Why the weekly is the right unit

Daily is too often for most agency work. Creative, strategy and content work doesn't move meaningfully in 24 hours, and a daily meeting becomes a status recital that nobody prepares for. Monthly is far too slow — a month is a quarter of a typical project, and by the time a monthly review notices a problem the budget is gone.

Weekly matches the actual rhythm of client work: approvals come back, weeks slip, invoices go out. It's also short enough that nobody can hide a bad number for long, and long enough that preparing for it isn't a burden.

The 45-minute agenda

Run it the same day and time every week. Attendees: whoever owns delivery, plus project leads. Not the whole agency. Time-boxed, and you keep to the boxes even when it's uncomfortable — the discipline is most of the value.

WEEKLY DELIVERY REVIEW — 45 minutes

00:00  (5 min)  The numbers
       For each active project: % of budget consumed vs % of work complete.
       No discussion yet. Just read them out.

00:05  (10 min) Red flags only
       Any project where budget consumed exceeds work complete by more
       than 10 points. Any project that missed a date this week. Any
       client who has gone quiet for more than five working days.
       Projects that are fine get zero airtime.

00:15  (15 min) The next two weeks
       Capacity by role for the next 14 days. Who is over-committed,
       who is under-committed, what starts, what finishes.
       Confirm every date that falls in the window has an owner.

00:30  (10 min) Client temperature
       One sentence per active client on how the relationship feels,
       from the person closest to it. Not the work — the relationship.
       Flag anyone whose tone has changed.

00:40  (5 min)  Decisions and owners
       Read back every decision made, with a name and a date.
       Anything without a name doesn't count as a decision.

END. If a topic needs more than its box, it becomes a separate
conversation with two named people, not an overrun of this meeting.

Copy that into a recurring calendar invite and put it in the description. The agenda travels with the meeting, so nobody has to remember it.

The one question that does the most work

Percentage of budget consumed versus percentage of work complete. It sounds obvious and it's still the single most useful number in agency delivery, because it catches trouble long before a deadline does.

A project at 62% of budget and 55% complete is fine — 7 points of drift is noise. A project at 80% of budget and 50% complete is in trouble right now, regardless of whether it has missed a single date, and you have half the timeline left in which to do something about it.

The reason this beats "are we on track?" is that "on track" is a feeling and people are optimistic. Two percentages are checkable, and the gap between them is either widening or it isn't.

Percentage complete does require someone to make a judgement call. That's fine. An honest estimate from the person doing the work, given weekly, is far more useful than a precise number derived from a task list that doesn't reflect the actual remaining effort.

Client temperature is not a soft metric

The ten minutes on relationship health tends to be the section people want to cut. Don't. Client relationships degrade in ways that never appear in a budget report: replies get shorter, a second stakeholder starts joining calls, approvals that used to take a day take four.

Those signals are known to exactly one person — whoever is closest to the client — and they don't get surfaced unless someone asks explicitly, every week, in a setting where "something feels off" is an acceptable answer.

Three questions that get real information out of people:

• Has their response time changed in either direction?
• Has anyone new joined the conversation?
• If they were going to complain about something next month, what would it be?

The third one is unreasonably effective. People know the answer.

Two rules that keep it from decaying

Projects that are fine get no airtime. The failure mode of every status meeting is round-robin reporting, where each person spends four minutes describing work that is proceeding normally. That's how a 45-minute meeting becomes 90 minutes and then gets cancelled for being useless. If the numbers are green and there's no flag, move on.

Every decision leaves with a name and a date. "We should talk to the client about the timeline" is not a decision. "Sam calls them Thursday about moving the launch to the 14th" is. Read them back at the end. It takes two minutes and it's the difference between a meeting that changes things and a meeting that describes things.

The monthly layer

The weekly handles active projects. Once a month, spend a longer session on the things that are too slow-moving for a weekly:

• Margin by project on everything that closed since the last one
• Utilisation by person over the month, and whether the pattern is sustainable
• Pipeline against the capacity plan for the next quarter
• Any renewal or scope conversation coming up in the next 60 days

Keep these separate. Mixing them into the weekly guarantees the weekly runs long and the strategic questions get five rushed minutes at the end.

Do this week

Put the 45-minute meeting in the calendar as a recurring invite, paste the agenda into the description, and run it once. Just once.

Before the first one, gather the budget-consumed and work-complete percentages for every active project. That's the hard part, and it's usually an hour of work the first time. Almost everyone who does it finds at least one project sitting at a gap they didn't know about — which is precisely the point of having a rhythm rather than a reaction.