The Coordination Tax

60% of knowledge work is spent on the mechanics of its own coordination — not on the work itself.

The 60% Problem

The Anatomy of Work Global Index, surveying over 10,000 knowledge workers globally, established that 60% of a knowledge worker’s day is consumed by “work about work” — communicating about tasks, searching for information, switching between applications, managing shifting priorities, and chasing status updates. Subsequent research has reinforced this finding: Microsoft’s 2024 Work Trend Index found that knowledge workers are now interrupted every two minutes during core work hours — roughly 275 times per day — by meetings, emails, or chat notifications, and spend an average of 11.3 hours per week (28% of the workweek) in meetings alone.8

Knowledge work breaks down into three categories: skilled work (the job you were hired to do), strategic work (planning and goal-setting), and coordination overhead (the interstitial tissue connecting them). The coordination layer has grown to be more voluminous than the activities themselves.

Category 2021 2022 Change
Coordination Overhead60%58%−2pp
Skilled Work26%33%+27%
Strategic Work14%9%−36%

The shift between 2021 and 2022 reveals a troubling trade-off: while workers spent 27% more time on skilled work, their time dedicated to strategy dropped by 36%. Reclaimed coordination time flows to execution, not to the long-term planning required to prevent future coordination crises.

The Thousand Tiny Cuts

The 60% is not the result of a single failure but a constant friction of digital fragmentation:

The trend is accelerating, not stabilising. Large meetings of 65 or more attendees are now the fastest-growing meeting type, and nearly a third of all meetings span multiple time zones — up 35% since 2021. Meanwhile, 57% of meetings are ad hoc calls without a calendar invite, and one in ten scheduled meetings are booked at the last minute.89

[Image needed: Coordination overhead visualization — pie chart of the 60/26/14 split or visual showing the thousand tiny cuts]

The Financial Cost

For a mid-sized organisation, the financial impact of coordination inefficiency can be quantified:

Cost Category Annual Cost (Est.)
Time Waste (15–20 hrs/week admin)$700K
Coordination Waste (20% productivity loss)$1.5M
Delayed Revenue (time-to-market gaps)$2.4M
Poor Decisions (20% error rate from bad data)$2.0M
Total Annual Coordination Tax$6.2M

Globally, PwC estimates over $3 trillion is lost each year to process friction. Across industries, 20–30% of operational expenditure is consumed by rework, miscommunication, and fragmented systems.

The Scaling Trap

Coordination overhead does not grow linearly with team size — it grows geometrically. Every new team member adds communication paths to every existing member. Doubling staff does not double output; it doubles coordination complexity.

This is Brooks’s Law applied to the modern knowledge economy: linear hiring cannot solve a geometric problem. Organisations that attempt to scale by adding headcount without addressing the coordination architecture create “diseconomies of scale” — the more people they add, the slower they move.

Organisations that address coordination overhead can reclaim three to four full-time equivalents of productive capacity in a ten-person team — without a single new hire.

Strategic Erosion

When coordination time is reclaimed, it does not flow to strategic thinking — it flows to more execution. Between 2021 and 2022, skilled work rose by 27% while strategic work fell by 36%. Organisations are doing more without thinking more.

This creates a reinforcing cycle: the lack of strategic clarity generates misalignment, which forces teams to spend even more time in the future realigning and correcting course. The coordination tax is self-perpetuating.

The AI Countermeasure

Generative AI and agentic automation are now the first structural counterforce to the coordination tax. McKinsey estimates that generative AI could automate activities that currently absorb 60–70% of employee time — up from the previous estimate of 50% for traditional automation — representing a potential $4.4 trillion annual productivity boost globally, roughly 4% of world GDP.10

The impact is already measurable. McKinsey’s 2025 State of AI report finds that nearly nine in ten organisations now regularly use AI in operations, and 75% of knowledge workers already use AI tools in some form — often without formal company deployment.11

AI agents — autonomous systems that can plan, execute, and coordinate multi-step workflows — are accelerating this shift. Gartner predicts that 40% of enterprise applications will feature task-specific AI agents by the end of 2026, up from less than 5% in 2025.12 By 2029, at least 50% of knowledge workers will develop new skills to work with, govern, or create AI agents on demand.12

Critically, the coordination layer is where AI delivers the most immediate value. Summarising meetings, triaging communications, drafting status updates, compiling reports, and routing approvals are precisely the “work about work” tasks that consume 60% of the knowledge worker’s day. The organisations that deploy AI against the coordination tax — rather than simply augmenting skilled work — will see the largest productivity gains.

The Compound Error Intersection

The coordination tax intersects directly with the compound error problem. The more agents coordinate, the more steps in the workflow, and the lower the cumulative success rate. At 99% per-step accuracy, a 100-step agent workflow succeeds only 36.6% of the time.

Consensus voting mitigates this by parallelising rather than sequencing: deploying multiple independent agents on the same input and using majority voting to reduce error. This architectural approach to reliability is fundamentally different from the coordination-heavy human approach of adding reviewers, approvers, and oversight committees to the chain.

The Implications for AI Governance

AI governance programmes are especially vulnerable to coordination overhead. They are, by design, coordination-intensive: risk committees, board presentations, approval chains, policy reviews, vendor assessments, incident reviews. These activities consume the thinnest, highest-judgment layer of human attention in the organisation.

If 60% of the people involved in AI governance are consumed by coordination overhead, the quality of governance decisions degrades. This is not a productivity problem — it is a governance quality problem.

The structural answer is not better project management. It is a fundamentally different operating model — one where AI handles the coordination and humans concentrate on the judgment that governance actually requires. With Gartner forecasting 40% of enterprise applications embedded with AI agents by 2026, the question for governance programmes is no longer whether to adopt agentic AI but how to govern the agents that are already arriving.12 Architecture-first governance moves enforcement into the system so that human attention is reserved for genuinely novel risk decisions, not routine approvals.

Sources

  1. Asana, “The Anatomy of Work Index 2021”
  2. Asana, “How Work About Work Gets in the Way of Real Work (2025)”
  3. HR Dive, “Drain of app switching: Why employees lose 5 hours per week”
  4. Crebos Online Solutions, “The True Cost of Operational Inefficiency”
  5. McKinsey, “The Economic Potential of Generative AI”
  6. Alyx Priestley, “The Product Ops ROI Calculator”
  7. Medium, “The Math That Kills Growth”
  8. Microsoft, “Work Trend Index: Breaking Down the Infinite Workday” (2024)
  9. Worklytics, “2025 Productivity Benchmarks for Knowledge Workers”
  10. McKinsey, “The Economic Potential of Generative AI: The Next Productivity Frontier” (2024)
  11. McKinsey, “The State of AI in 2025”
  12. Gartner, “Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026” (2025)

See How Architecture-First Governance Addresses the Coordination Problem

Our methodology moves governance enforcement into the system — so human attention is reserved for the decisions that matter.

Schedule a Briefing Augmentation & Digital Apprentices