Somewhere between the arrival of always-on chat and the normalisation of back-to-back video calls, knowledge work acquired a measurement problem. Organisations could see activity in extraordinary detail — messages sent, meetings attended, documents touched, hours logged in a platform — and almost none of it told them whether anything valuable had been produced. The uncomfortable finding, repeated across research and internal studies, is that the two are frequently inversely related. The people generating the most visible activity are often the least able to complete work requiring sustained concentration, because visible activity is made of interruptions, and interruptions are expensive in a way that no activity dashboard captures. The cost is not the interruption itself. Research on workplace attention has consistently found that returning to an interrupted task takes far longer than the interruption lasted, and that people compensate for fragmentation by working faster — at the price of higher stress and frustration. A calendar with six short meetings scattered across a day does not leave six hours of work time. It leaves a series of gaps too short to start anything that requires holding a problem in your head.
Why the tooling made it worse
Collaboration platforms did not create interruption, but they industrialised it, and three design choices did most of the damage. Presence indicators turned availability into a performance. A green dot is a claim about responsiveness, and once colleagues can see it, being seen as available becomes part of the job. The predictable behaviour is to keep the app open, keep responding, and do the concentrated work at the edges of the day. Default-on notifications made every message urgent. A channel post and a production incident arrive with the same signal. Absent deliberate configuration, the platform treats all information as equally interrupting, and most organisations never configure it. Frictionless meeting scheduling removed the cost of asking. When booking an hour of six people's time takes ten seconds and no justification, the volume goes up. Calendar systems that show free slots optimise for filling them. None of these are defects. They are the product working as designed, and the design assumed responsiveness was the goal. Where responsiveness genuinely is the goal — support, operations, incident response — the tools are excellent. Where the goal is analysis, writing, design or engineering, the same defaults are actively hostile.
What the organisations that fixed it actually did
The interventions that worked share a characteristic: they changed defaults and structures rather than asking individuals to have better willpower. Meeting-free blocks at the organisational level. Not an individual's blocked calendar, which anyone can override, but a protected period across a team or company. Half a day, twice a week, is the most common pattern and the most durable. Explicit response-time expectations by channel. Chat within the working day rather than within minutes. Email within a day. A named path — phone, page, a specific channel — for genuine urgency. Most of the pressure to respond instantly comes from ambiguity about whether instant responses are expected. Written asynchronous updates replacing status meetings. The recurring status meeting is the highest-volume, lowest-value interruption in most organisations, and a written update read when convenient is strictly better for everyone who is not running the meeting. Defaults changed centrally. Notifications off outside working hours, meetings defaulting to 25 and 50 minutes, agendas required, and a standing permission to decline meetings without a stated purpose. Defaults do more work than policies. Leaders modelling it visibly. An executive who sends messages at midnight while saying they do not expect replies has communicated the opposite of what they said. This is the intervention most often skipped and most often the reason the rest fails.
| Visible activity | What to examine instead |
|---|---|
| Message volume | Whether channels have explicit response-time expectations |
| Presence time | Whether focus time and unavailability are permitted |
| Calendar occupancy | Meeting load, fragmentation and stated purpose |
| Status-meeting attendance | Whether a written update could meet the same need |
Qualitative summary of this article's source text, not a measured outcome or performance estimate.
Practical Guidance for Collaboration Health Review
- Measure meeting load and fragmentation, not message volume. Count uninterrupted blocks of two hours or more per person per week; that number is the one that predicts whether deep work happens.
- Protect focus time structurally, not individually. A company-wide or team-wide meeting-free block survives; a personal calendar block does not.
- Publish response-time expectations per channel. Ambiguity, not urgency, is what keeps people in the app.
- Change platform defaults centrally. Quiet hours, shorter default meetings, notification settings and required agendas do more than any awareness campaign.
- Replace recurring status meetings with written updates. Keep the meeting only where discussion or decision is genuinely required.
- Give explicit permission to decline and to be unavailable. Then have senior people use it visibly, which is the only version that works.
- Audit who is in the room. Optional attendees who feel obliged to attend are the cheapest saving available in most organisations.
- Never use activity metrics for performance management. The moment green-dot time or message counts affect evaluation, you have optimised for the wrong thing permanently.
The Regional Dimension
The Gulf adds structural complexity to this problem that generic guidance does not address. The working week is not uniform. Weekend days differ across countries in the region, and within a single group an entity in one country may be at rest while another is mid-week. For teams distributed across the GCC, the overlap window is compressed, and the practical consequence is that everyone's meetings are pushed into the same narrow band — which is exactly the condition that destroys concentration blocks. Groups operating across several regional countries plus Europe or Asia find the workable overlap is a few hours, and that those few hours end up wall-to-wall. Ramadan changes working hours materially, with legally reduced hours in several jurisdictions and a significantly altered daily rhythm. Organisations that continue running the same meeting volume in a shortened day have simply removed all the remaining focus time. The better practice — and it is worth committing to in advance rather than deciding each year — is to cut recurring meetings substantially for the period and shift to written updates, with decisions batched into fewer, purposeful sessions. The hierarchical norm in many regional organisations matters too. Where responsiveness to senior leadership is a strong cultural expectation, a message from a manager is not something an employee feels able to defer, whatever the written policy says about response times. Publishing expectations therefore does very little unless senior people actively demonstrate that deferred responses are acceptable — and in practice that means a leader saying "answer this tomorrow" in the message itself, repeatedly, until it is believed. Two further local factors. Multilingual teams carry an additional cognitive load that rarely appears in these discussions: working across Arabic and English, and often several other languages, means more time spent processing communication, and fragmentation costs more for someone working in their second or third language. And the dominance of messaging apps in regional business communication means the interruption surface extends onto personal phones outside any corporate control — a WhatsApp group with a chairman in it is not a channel where anyone sets a notification schedule.
The objection worth taking seriously
The deep work argument has a real weakness: it is largely an argument made by and for a specific kind of worker, and it is often used to justify unavailability by people whose colleagues cannot afford the same. Much of the world's work is legitimately interrupt-driven. Customer support, clinical care, operations, logistics, incident response, retail and hospitality are responsive by definition, and telling those teams to protect focus blocks is not advice, it is noise. Even within knowledge work, the person protecting four uninterrupted hours is frequently doing so at the cost of a colleague who is now blocked waiting for an answer. Focus time is not free; it moves the cost onto whoever needs the response. The honest framing is a trade-off between individual throughput and collective latency, and the right balance differs by role and by team. There is also a fair objection that the whole conversation individualises a structural problem. Most interruption in an organisation is a symptom of something else: unclear ownership, so everything requires consultation; poor documentation, so people ask rather than look; over-broad approval requirements; and teams whose dependencies are badly drawn. Telling people to manage their notifications while leaving those conditions in place treats the symptom. The organisations that genuinely reduced interruption did it by clarifying who decides what. And a measurement caution: this field has produced a great deal of confident quantification of the cost of interruption, much of it extrapolated well beyond what the underlying studies support. The direction of the finding is robust — fragmentation is costly, and recovery takes longer than people assume. The precise numbers in circulation should be treated with more scepticism than they usually receive.
Common Questions
What should we actually measure?
Uninterrupted blocks of two or more hours per person per week, meeting hours as a share of the working week, and the proportion of meetings with a stated purpose. Avoid message counts, presence time and any metric that rewards visible activity.
Do meeting-free days work?
Protected blocks across a whole team or organisation work. Individually blocked calendars do not, because they get overridden. Half a day twice a week is more sustainable than a full day, which tends to displace meetings into adjacent days at higher density.
How do we handle genuinely urgent issues if people are not responsive?
Define one escalation path that is always answered — a phone call, a paging system, a named channel — and keep it narrow enough to remain credible. Most organisations discover that true urgency is a small fraction of what currently gets treated as urgent.
Does AI reduce or increase the problem?
Both, and the net effect depends on a governance choice. It genuinely reduces some load: meeting transcripts and summaries make attendance optional for information-only sessions, thread summarisation removes the need to read everything, and drafting assistance shortens the writing that follows a decision. That is real recovered time. But the same tools make it dramatically cheaper to produce communication, and output that costs nothing to generate tends to get generated — longer documents, more updates, more messages, all of which someone still has to absorb. There is also a new failure mode: if every meeting is recorded and summarised, the marginal cost of scheduling one falls further, and meeting volume rises. The organisations that benefit will be the ones that use AI to remove meetings and reading, not to make producing more of both effortless.
Collaboration Health Review — measure uninterrupted hours, change defaults centrally, and remember that focus time protected by one person is usually paid for by another.