Burnout's new leading indicators: cognitive strain and decision friction

For most of its history, burnout has been measured in hours. The longer the workload, the higher the risk; the fix was to lighten the load. That model is now breaking down. Deloitte's 2025 Workforce Intelligence Report found that mental fatigue, cognitive strain and decision friction have overtaken workload volume as the leading indicators of burnout, for the first time on record.¹ The thing to watch is no longer how much work people carry, but how mentally expensive each hour of it has become.

The shift matters because it changes what counts as a warning sign, and most organisations are still reading the old ones. Eagle Hill Consulting found that 55% of the US workforce reports burnout, and framed it bluntly as a performance problem rather than a wellbeing footnote: a drag on efficiency, innovation, customer service and retention.² When a majority of people are running hot and the damage lands on output and attrition, burnout belongs on the operating-model agenda, not only the wellness one.

Treat the precise causal ranking as a direction of travel rather than settled science; it reaches us through vendor research relayed secondhand. The structural evidence underneath it is harder to wave away.

Why hours-worked metrics arrive late

The reason a workload metric misses modern burnout is that the strain has moved into the texture of the day. Microsoft's aggregated Microsoft 365 telemetry shows employees interrupted every two minutes during core hours, roughly 275 times a day, with nearly half of employees (48%) and more than half of leaders (52%) describing their work as chaotic and fragmented.³ None of that registers on a timesheet. A person can work a contained week and still finish it cognitively shattered, because the cost was paid in fragmentation, not duration.

Decision friction is the name for that hidden cost: the cumulative tax of moving between scattered tools, ad hoc meetings and unclear ownership. Aggregated research suggests employees lose around three hours a day simply searching for information across fragmented systems, and that 47% report feeling burned out or frustrated when they lack the tools or information to do their jobs, a figure up twelve points year on year.⁴ Those numbers come from vendor and aggregated reporting and should be read as colour rather than proof, but they track with what the telemetry shows. The toolchain itself has become a wellbeing variable.

A structural layer sits beneath even that. The OECD finds that roughly one in three job vacancies across member economies now carries high AI exposure, reshaping the tasks those roles involve and the skills they demand.⁵ Roles defined for a pre-AI workflow now sit on top of an AI-shaped, fragmented reality, and the distance between the two is itself a generator of friction. The lever here is role redesign, not another resilience webinar.

The trap of watching people to protect them

If the signals worth tracking are behavioural, the obvious move is to start tracking behaviour. This is where good intentions tip into harm. The American Psychological Association reports that close behavioural monitoring is, in the researchers' words, extremely stressful: it limits autonomy and breeds fears of job insecurity, with 56% of monitored workers feeling tense or stressed at work against 40% of the unmonitored.⁶ Since stress and insecurity are upstream of burnout, instrumenting people to catch their burnout early risks adding to the strain it claims to detect.

The objection deserves to be met squarely rather than dodged, because it is right as far as it goes. The harm in that evidence attaches to covert, individual-level watching where the worker is the subject. The resolution is architectural, not rhetorical. There is a difference between watching individuals, which adds strain, and surfacing structural patterns to the people who own them, which relieves it. An early-warning model built the second way is employee-consented and employee-visible: the individual sees their own signals first, and consent to share flows upward, never the reverse.

That distinction also separates a precursor from a consequence. Tools that score activity and tally hours detect burnout downstream, after it has already surfaced in falling output or a resignation letter, and they read it through data employees have every reason to game. A model that watches structural leading indicators catches them before they harden into attrition. The signal and the ethics point the same way, which is rare enough to be worth saying plainly.

A three-signal model

The practical version reduces to three families of signal, each surfaced with consent. Start with cognitive-load proxies — interruption cadence and fragmentation density, the 275-a-day reality made visible per team. Alongside them sit temporal patterns: after-hours and weekend creep, measured against a person's own baseline rather than a uniform threshold. The last family is role-fit, the mismatch between how a role is defined and how its digital work actually happens — the OECD gap brought down to the level of an individual job.

None of these is a licence to watch harder. Each is a prompt to fix the structure, by consolidating tools, clarifying ownership, protecting deep-work time and redesigning roles that no longer fit their workflow. The aim is to relieve the strain, not to instrument the suffering and call it insight.

So the real question for 2026 is not whether an organisation can measure burnout. It is whether leaders are prepared to act on the precursors and redesign the work itself, or content to keep counting the casualties after the fact. Most, on current evidence, are still counting.

Footnotes

1. HRD Connect. (2025, December 4). Burnout is back: How organisations can reset for a healthier, more sustainable 2026 (citing Deloitte Workforce Intelligence Report, 2025). https://www.hrdconnect.com/2025/12/04/burnout-is-back-how-organisations-can-reset-for-a-healthier-more-sustainable-2026/ ↩

2. Eagle Hill Consulting. (2025). Workforce burnout survey 2025. Eagle Hill Consulting. https://www.eaglehillconsulting.com/news/workforce-burnout-survey-2025/ ↩

3. Microsoft. (2025). Breaking down the infinite workday (Work Trend Index Special Report). Microsoft WorkLab. https://www.microsoft.com/en-us/worklab/work-trend-index/breaking-down-infinite-workday ↩

4. Speakwise. (2026). Cognitive load statistics 2026 (citing Coveo 2025 Employee Experience Relevance Report). Speakwise. https://speakwiseapp.com/blog/cognitive-load-statistics ↩

5. OECD. (2025). Bridging the AI skills gap: Is training keeping up? OECD Publishing, Paris. https://www.oecd.org/content/dam/oecd/en/publications/reports/2025/04/bridging-the-ai-skills-gap_b43c7c4a/66d0702e-en.pdf ↩

6. American Psychological Association. (2024). Electronically monitoring your employees? It's impacting their mental health. APA Healthy Workplaces. https://www.apa.org/topics/healthy-workplaces/employee-electronic-monitoring ↩