A late-night availability indicator is easy to see. Valuable work is harder to evaluate. When leaders substitute visible activity for results, they risk rewarding endurance while overlooking quality, judgment, innovation, and the health of the people doing the work.
The useful distinction is simple: hours are an input, not an outcome. High-performing organizations manage valuable output, quality, and sustainable capacity together. Persistent overtime deserves investigation, not automatic praise. It may signal unclear priorities, excessive meetings, understaffing, fragile systems, or work that should have been stopped.
This is not an argument that nobody should ever work late. Product launches, incidents, seasonal demand, and genuine emergencies sometimes require a surge. The question is whether long days are an exception with recovery afterward or the normal operating model.
Separate activity from productivity
The US Bureau of Labor Statistics describes labor productivity in terms of how efficiently inputs are converted into outputs. Its productivity program compares growth in output with growth in hours worked. By definition, adding hours is not the same thing as improving productivity.
If a team works 20 percent more hours to produce 10 percent more comparable output, total output rose while output per hour fell. If output and hours rise proportionally, productivity is unchanged. Longer hours can coexist with higher productivity, but the hours do not prove it.
The numerator needs care, too. A support team can close more tickets by rushing customers off calls. A software team can ship more features while creating defects and technical debt. A sales team can book more meetings without improving revenue or retention. Useful measurement combines throughput with quality and outcomes.
Readiness also has value. An operations team cannot be judged only by the number of incidents it handles. Quiet periods can reflect good preventive work, and available capacity can be essential insurance. The goal is not to squeeze visible activity from every minute. It is to create reliable value over time.
What the evidence actually supports
The International Labour Organization published Working Time and Work-Life Balance Around the World on January 6, 2023. It reported that more than one-third of workers globally regularly worked more than 48 hours per week.
The ILO’s accompanying summary says longer hours are generally associated with lower unit labor productivity, while shorter hours are associated with higher productivity. It also identifies potential organizational benefits from flexible work arrangements and costs from restricting them, including increased turnover.
That evidence does not prove that shortening any particular team’s schedule will automatically improve its results. Industry, job design, staffing, worker preferences, and customer needs all matter. It does support treating chronic overwork as a management question rather than a cultural virtue.
Health evidence makes that question more urgent. In a joint release dated May 17, 2021, the World Health Organization and ILO estimated that long working hours accounted for 745,000 deaths from stroke and ischemic heart disease in 2016, 29 percent more than in 2000.
Their analysis associated working at least 55 hours weekly with a 35 percent higher risk of stroke and a 17 percent higher risk of dying from ischemic heart disease compared with working 35 to 40 hours. These are population-level associations, not predictions for an individual. They are nevertheless strong reasons not to build a company culture around sustained 55-hour weeks.
The infinite workday is often fragmented work
Microsoft examined aggregated and anonymized Microsoft 365 activity to understand how communication fills the modern workday. Its June 2025 report, Breaking down the infinite workday, found that meetings, email, and chat frequently compete with focused work. Among employees with the highest volume of incoming notifications, a meeting invitation, email, or chat arrived an average of once every two minutes during an eight-hour workday.
That finding does not mean every employee is interrupted every two minutes. It applies to the 20 percent of users who received the most notifications, and it measures incoming activity rather than the amount of productive time lost. Even with those limits, the data illustrates how a long workday can still leave little uninterrupted time for meaningful work.
The practical point is still useful. A person can be online for 12 hours while spending much of that time switching contexts, sitting in low-value meetings, searching for information, and responding to other people’s priorities. Extending the day does not repair a poorly designed day.
Leaders should examine which coordination is necessary and which is organizational noise. Protecting focus time does not mean eliminating collaboration. It means making collaboration intentional enough that people can also think, build, decide, and finish.
Recovery is productive infrastructure
Time away from work is often discussed as a benefit exchanged for labor. It is also part of the system that makes sustained performance possible. Rested employees can return with more attention, patience, perspective, and capacity for complex judgment.
That does not make every vacation policy effective. Leave that exists on paper but cannot be scheduled does little good. A culture where senior leaders send routine weekend messages can undermine formal policies even when they say no reply is required. Teams need enough coverage, documentation, cross-training, and managerial trust for people to disconnect without work piling up behind them.
Managers should therefore treat unused leave, repeated weekend work, concentrated on-call load, and dependence on individual experts as operational signals. None proves burnout by itself. Together, they can reveal a system with too little resilience.
A better organizational scorecard
No single metric can replace hours. The answer is a balanced set of measures appropriate to the work.
1. Valuable outcomes. Start with what customers, users, or the organization actually need: resolved problems, retained customers, successful transactions, reliable products, better decisions, or reduced risk. Count completed activities only when they are reasonable proxies for value.
2. Flow and cycle time. Measure how long comparable work takes from request to completion. Look at both typical duration and variation. Faster is useful only when quality and outcomes remain acceptable.
3. Quality and rework. Track defects, reopened cases, failed changes, returns, complaints, and effort spent correcting prior work. Rising rework alongside overtime is a reason to investigate, though it does not prove fatigue caused the problem.
4. Reliability and service levels. Measure whether products and services meet agreed expectations. Include availability, response times, recovery performance, and completion of preparedness exercises where relevant.
5. Sustainable throughput. Review completed work over a rolling period alongside overtime, after-hours activity, leave taken, staffing changes, and work complexity. A short surge may be healthy. A permanent surge is simply the operating model.
6. Customer and employee signals. Use customer satisfaction, retention, voluntary turnover, vacancies, and a small set of consistent team-health questions. Treat these as prompts for investigation rather than verdicts about individuals.
Balance the measures to reduce gaming. Speed without quality encourages rushed work. Volume without complexity rewards easy tasks. Satisfaction without cost can hide an unsustainable service model. Keep definitions stable, document changes in demand and staffing, and review conflicting signals rather than selecting the most flattering number.
Do not turn the scorecard into surveillance. Keystroke counts, webcam monitoring, online indicators, and individual message totals are easy to collect but weak evidence of value. Intrusive measurement can also damage the trust and autonomy that knowledge work requires.
Where AI fits
Microsoft’s 2026 Work Trend Index annual report, published May 5, 2026, surveyed 20,000 AI-using knowledge workers across 10 markets. It reports that 66 percent said AI allowed more time for high-value work, while 86 percent treated AI output as a starting point rather than a final answer. These are self-reports, not measured time savings or guaranteed business results.
AI and automation should remove low-value work, improve decisions, or expand useful capacity. They should not become an excuse to fill every reclaimed minute with more tasks. Measure the full workflow, including review and correction, and ask whether the change improves outcomes without unacceptable quality, security, cost, or workload tradeoffs.
Why This Matters for Government
The same principles apply to public organizations, where productivity must include public value. A faster eligibility decision is not an improvement if it is incorrect or inaccessible. An IT team that ships quickly but leaves critical services fragile has not delivered a good outcome.
Government leaders also manage emergency coverage, constrained staffing, and stewardship of taxpayer money. Routine workload planning should preserve surge capacity, backup coverage, handoffs, and recovery after demanding incidents. Repeated overtime should trigger a review of priorities, staffing, dependencies, and recurring failures, not serve as the primary proof of commitment.
The leadership question
Replacing an hours culture starts with a better question. Instead of asking, “How long were people online?” ask, “What valuable result did we create, at what quality, and can we do it again without exhausting the team?”
Hours still matter for staffing, compensation, coverage, and legal compliance. They belong in the scorecard as inputs and constraints. They do not belong there as substitutes for performance.