The real cost of context-switching for agency teams

There is a specific kind of Tuesday in an agency that goes like this. You sit down at 9:02 with a coffee and an honest plan: two hours on the rebrand deck before the 11am stand-up. At 9:07 a Slack notification lands from the e-commerce client asking why a tracking pixel "looks weird". At 9:14 your finance lead pings about an unsigned SOW. At 9:21 a designer DMs the wrong Figma link. At 9:28 you remember the rebrand deck. By 9:35 you have opened Notion, Asana, Harvest, Slack, Gmail and a Loom you do not remember being sent. You have not made the deck. You have made yourself a second coffee.

That morning is the job. It is also, on closer inspection, a small financial event with consequences.

The research everyone cites - and why it underplays agency life

The numbers on context-switching are well-rehearsed at this point. Gloria Mark's research at UC Irvine found that knowledge workers need an average of 23 minutes and 15 seconds to fully refocus after a single interruption. They rarely return to the original task directly, with usually two intervening tasks in between. The American Psychological Association's switch-cost research, conducted by Meyer, Evans and Rubinstein, found that frequent task-switching can consume up to 40% of someone's productive time. A 2022 study cited by Harvard Business Review tracked the average digital worker toggling between apps and websites nearly 1,200 times per day. RescueTime puts the average switch interval at roughly 40 seconds.

McKinsey's research on knowledge work found the average worker spends close to 60% of the week, around 28 hours, on email, hunting for information and internal coordination rather than the core job. Agencies sit at the heavier end of that distribution because multi-client work compounds the load.

That is the well-rehearsed canon. It is also, for agency teams, an undercount.

Why agencies are worse than knowledge work in general

Most context-switching research treats one knowledge worker juggling multiple tasks inside one organisation. Agencies do not work that way. An agency staffer juggles tasks inside multiple organisations. Each client is its own little firm with its own brand voice, brief history, approval chain, budget burn, preferred channel and Slack-vs-email politics.

Switching from Client A to Client B is not switching tabs. It is changing employers for forty-five minutes, then changing back. You reload tone of voice, deliverable status, who-said-what-on-Thursday's-call, where the asset is hiding, which Notion page is canonical and which Asana board has the live brief. The cognitive lift on each switch is heavier because the contexts are more distant from each other.

The tool stack compounds it. Most studios we have worked with are running some combination of Asana or ClickUp for project work, Notion for docs, Slack for chat, Harvest or Toggl for time, Figma for design, a CRM for sales, Xero or QuickBooks for finance and a folder structure in Google Drive that nobody trusts. Each client lives slightly differently inside each one. A Slack workspace shared with Client A, a guest channel with Client B, a shared Notion with Client C, email-only with Client D. None of it talks to the others.

Senior people often end up as the integration layer between systems that refuse to integrate. That is the role nobody hired them to do.

The hidden costs nobody puts on the P&L

The first cost is utilisation. The APA's 40% figure measures productive capacity across general knowledge work, not billable agency hours directly. Translating: if a senior team member with seven hours of available billable time loses something like 40% of their productive capacity to switching, you are not getting seven billable hours out of them. You are getting closer to four. Multiply that across a team of fifteen and the gap between "available capacity" and "real capacity" becomes a likely reason your forecast keeps missing.

The second is billable-hour leakage. Switching time is rarely captured properly on timesheets. People log "client work" against the visible blocks and quietly absorb the cognitive ramp-in around them. Industry estimates of uninvoiced billable hours vary, with commonly cited figures around 10 to 20%. We would argue switching tax accounts for a meaningful share of that, though no one has isolated the figure cleanly.

The third is senior-team fatigue. A 2024 Mentally-Healthy Survey of media, marketing and creative professionals across the UK, US, Australia and New Zealand found that 70% had experienced burnout in the previous 12 months, a meaningful step above the broader knowledge-work average. Constant context-switching is not the only driver, but it is one of the most consistent ones operators name. The brain does not get a clean break between clients. It gets a low-grade alarm running all day.

The fourth is client-experience drift. The work itself starts to thin out. A designer who switches between three brand systems before lunch is producing competent work for all of them, but the sharp work (the bit you charge premium for) only happens after the warm-up cost of full immersion. Switching denies that warm-up almost every day.

Put it in numbers, with the caveat that these are worked examples rather than measured figures. Take a senior designer or strategist on a fully-loaded cost of around $120 an hour (use your own number). Apply the APA's 40% switching tax to a 35-hour week. That is 14 hours, or roughly $1,680 a week, $80,000 a year of capacity that does not reach an invoice. Across three or four senior people in a 15-person studio, the back-of-envelope figure clears $250,000 a year. The actual number for any given agency depends on rate, utilisation and how switching-heavy each role really is, but the order of magnitude is consistent with what most operators we have spoken to estimate.

What separates the agencies that solve it

The agencies we have seen get this under control do not solve it by demanding focus alone. Telling the team to focus harder rarely works on its own when the structure pulls in the opposite direction.

They change the structure in three specific ways.

They collapse the tool stack toward one operating model. Not necessarily one tool, but one source of truth for project status, time, scope and margin. When a senior person can answer "is this project still profitable" without opening three browser tabs and a spreadsheet, the integration burden disappears off their plate. AgencyFlo exists because we needed exactly this and could not buy it.

They define response windows instead of pretending to be always-on. Slack is checked in batches (9:30, 12:30, 4:00), not in real time. Client expectations are set explicitly in the SOW: "We respond to non-urgent messages within four working hours. Urgent issues use the phone." In our experience, few clients churn over a four-hour response window, provided the expectation is set upfront.

They batch comms work into a daily ops slot. Status updates, internal pings, "quick questions": all of it gets a slot, usually morning and end-of-day. The rest of the day is for the work the agency is actually paid for. The "no Slack before 11" rule sounds glib until you watch a designer ship two days of work in one morning.

None of these patterns are novel. Many of the agencies we have seen improve margin run some version of them.

The one move to make this week

Pick the next tool-stack split that costs you the most context-switching (usually time-tracking sitting in one app and project status sitting in another) and consolidate one of them. If you cannot consolidate this quarter, at minimum stop maintaining both. Pick the system of record and let the other one go stale. Tell the team explicitly which one is canonical.

If you want a sharper test, run a one-week single-app trial. For five working days, the team works out of the smallest possible set of tools (ideally one project surface, one chat surface, one doc surface) and any new tool requires a written case. Track the work that ships and ask the team how the week felt. Most agencies that run this find they were carrying two or three tools they could quietly remove. The week after, they remove them.

Context-switching is not only a discipline problem. It is also an architecture problem - and that is where most of the gains are still on the table. The agencies that treat it that way get measurable senior hours back. The ones that treat it as pure discipline tend to keep wondering why margin is thin and the team looks tired.