A different post from the four-day-week-as-hiring-strategy argument. This is the actual operational model that makes a four-day week work in services. The part where, in conference Q&As, agency leaders genuinely want the practical detail and most published case studies don't go there.
We've operated this way since 2022. Some of what follows we got right early. Some of it took two years to figure out. All of it is specific to how a small-to-mid digital agency actually runs.
The default: async-first
The single most important operational decision is that work runs async by default and synchronous only when the synchronous mode adds something specific.
This means:
Most strategic work happens in writing. A new campaign brief, a quarterly plan, a client onboarding document — these are produced as written artefacts that the relevant people read, comment on, edit, and approve, asynchronously. The meeting that, in many agencies, would be the venue for these decisions, doesn't happen. The decision is made in the document.
Meetings are reserved for genuinely synchronous work. (How meetings die under a four-day week covers the specific cancellation rules.) Three categories: relationship work (client calls, new business pitches, internal one-to-ones), creative collaboration (brainstorms where the back-and-forth genuinely needs to be live), and decisions that require multiple people in the same conversation at the same time (most of which don't, on inspection, actually require that).
Documentation is the unit of organisation. Most teams say they want better documentation. Few are willing to invest the time to produce it. We've made the investment. The result is that new joiners can come up to speed substantially faster than they could in our previous model, because the operational knowledge is captured in writing rather than in senior heads.
The async-first default isn't trivial to set up. It requires that the senior team writes well, that the document templates are good, that the comment-and-resolve cycle is genuinely used, and that the team actually believes the decisions are being made in the documents rather than in some shadow conversation outside them.
In our first year on the four-day week, we had to push back hard against the natural tendency to schedule meetings whenever something felt important. Year three, the team mostly polices itself. The cultural norm has stabilised.
Capacity planning when one day is gone
The straightforward issue with a four-day week is that the team has 20% less time. The unstraight-forward issue is that the way capacity is usually planned in agencies — by allocating people to projects in person-day units — needs a structural rethink.
Two changes that mattered.
We plan capacity in committed hours, not headcount-days. A senior strategist might be allocated 24 hours to a client engagement over a fortnight, not "three days" in the abstract. The shift to hours forces clarity about the actual work needed and about how much of the senior person's time is genuinely committed vs notionally allocated.
We over-allocate at the senior level and under-allocate at the junior level. Senior staff are 90-95% utilised against client work. Junior staff are 70-75% utilised. The ratio is the inverse of what most agencies run. The reason: senior time is genuinely scarce in our model, and protecting it from internal admin and meetings is the operational priority. Junior time is more easily flexed and the buffer absorbs the scheduling chaos that's inevitable in services.
This works because the agentic systems we've built (covered in another post) absorb a meaningful share of the workflow tax that used to be junior work. The junior layer is smaller relative to senior than it would be in a traditional agency, and the junior layer's work is more thinking-heavy than process-heavy.
Meeting rules
The specific rules that, in practice, hold meetings under control.
No standing meeting that hasn't justified itself in the last quarter. Standing meetings drift toward irrelevance over time. We review the meeting calendar quarterly and kill the ones whose purpose has eroded. Most kept meetings re-justify themselves. The ones that don't, go.
Meetings have agendas, or they're not meetings. A meeting without an agenda gets cancelled or converted to a doc. The agenda is the entry ticket. People who turn up to a meeting without one have demonstrated they don't have a specific reason to be in the meeting, which is a signal they probably shouldn't be there.
Meetings end with documented decisions and owners. Not optional. The output of every meeting is a writeup with the decisions that were made and who's responsible for them. Most decisions don't need a meeting; meetings that produce no decisions are clearly not needed at all.
No internal meeting longer than 30 minutes without a specific reason. Default duration is 30 minutes. Anything longer requires justification at the point of scheduling. Almost all meetings finish in under 30. The 60-minute default that most calendar apps set is a major cause of unnecessary meeting time.
These rules sound restrictive. In practice they're freeing. The team has substantially more uninterrupted time to do actual work. The meetings that do happen are more useful.
Capacity planning around the lost day
The structural decision: which day do we not work?
We chose Friday. Several reasons. Client communication tapers naturally on Fridays. Internal urgency tapers as people mentally close out the week. The day before the weekend already has the lowest output across most services teams, even those operating five days. Removing it as a working day codified a pattern that was already partly there.
The trade-off: client urgency that lands late Thursday afternoon doesn't get answered until Monday. We've found, contrary to early fears, that this is rarely actually a problem. Genuinely urgent client issues — the kind that can't wait two business days — are rarer than people anticipate. We have an on-call rotation for genuine emergencies that's used roughly twice a year. The rest of the perceived urgency turns out to be standard work that can wait until Monday morning.
For some services businesses, Wednesday or Monday might be a better choice. The argument for not-Friday assumes a Western corporate calendar where Friday is naturally lower-tempo. If your customers are in a rhythm where Fridays are peak, this won't apply.
What we automated to make this possible
The four-day week works in our model because we've offset enough of the internal workflow tax with agentic systems that the senior team genuinely has the capacity to do client work in 32 hours that used to take 40.
The four workflows that mattered most: audit pipelines, weekly reporting, proposal generation, QA on outbound work. (Detailed in another post.) Together they recover roughly 400 hours a month of senior time, which is the operational equivalent of having an extra senior person on the team — except without the management overhead.
Without the agentic layer, the four-day week would still be possible but the maths would be tighter. With it, the four-day week becomes operationally comfortable rather than operationally stressed.
For agencies considering the move without a robust agentic infrastructure, the realistic version is one where margin pressure is higher in year one and you're depending more on the hiring filter (senior, output-oriented, self-directed people) to make the model work. Doable, but the path is harder than it is for an agency that's done the operational work first.
The client communication piece
The thing most agency leaders worry about is client perception. Will clients be unhappy about reduced availability. Will they see it as a reduction in service. Will they want a discount.
Three years in, the answer is overwhelmingly no.
We're explicit with clients during onboarding that we operate four days a week. We position it as a productivity strategy, not a perk. We commit to specific response times within the four-day week (24-hour response on Monday-Thursday, Monday morning response for anything that lands late Thursday or Friday). We deliver against these commitments.
Clients, in our experience, care about responsiveness within the working week and quality of outcome. They care less about which days the working week falls on. We've had two clients in three years express any concern about the four-day model. Neither was a serious objection — both were satisfied by the response-time commitment in the contract.
The model probably wouldn't work for an agency whose service is, at the core, on-call support. For most strategic and execution-led services, it's a non-issue.
Closing
The four-day week works in services if the operating model is rebuilt around it. It doesn't work as a perk bolted onto a standard agency operating model.
The pieces that matter, in approximate priority: async-first defaults, meeting discipline, hour-based capacity planning, agentic offloading of internal workflow tax, and clear client communication about response commitments.
Most of these are independently good practice. Operating a four-day week forces the discipline of actually doing them, because the maths doesn't work otherwise. That forcing function is part of why the four-day week, done seriously, ends up improving the operational quality of the agency rather than just shifting where the working week happens.
If you're considering the move and want to talk through the operational specifics, drop a note. The agency-leader-to-agency-leader conversation tends to be more useful than the published versions of this material.