Five-Day Office Attendance Hits 67%—Space Data Is Still Missing

On August 27, 2026, Johnson Controls published its 2026 workplace survey findings, putting five-day office attendance at 67% among employees at the large U.S. organizations represented—up from 57% in 2025 and 48% in 2024. Among respondents using workplace-management technology, 75% had space-planning tools, 67% workplace analytics and 45% environmental or occupancy sensors; 65% of business leaders used AI in workplace operations.

The operational message is less reassuring than the attendance headline. A more predictable office week can provide a planning baseline, but it does not reveal which desks, meeting rooms or floors are used, for how long, or whether the existing footprint is efficient.

Attendance is stabilizing, but the benchmarks measure different things

The headline percentage describes work patterns provided by business leaders. It is not a continuous count of people inside buildings, a direct employee poll or a measured utilization rate, and the organizations represented should not be treated as a proxy for every U.S. workplace or other English-speaking markets.

A separate benchmark points in the same broad direction while measuring a different subject. The 2026 CBRE Americas occupier survey, published July 30, found that 89% of employers required at least three office days, up from 78% in 2025, while actual attendance averaged 2.9 days against a 3.2-day target.

These findings do not establish a single market-wide attendance rate. The Johnson Controls benchmark classifies reported employee work patterns, whereas the CBRE research compares attendance requirements with average attendance. Together, they indicate that office expectations are settling even though presence still varies by organization and can remain below policy targets.

Entry data cannot show whether individual spaces are working

Attendance, occupancy and utilization answer different questions. Attendance describes whether or how often people come to the workplace. Occupancy measures how many people are present in a building or zone at a given time; utilization measures whether a particular desk, room or area is used and for how long.

An access-control event can establish that a credential crossed an entrance, but it cannot identify where its holder went inside. Reservation systems capture intended demand, yet a booked room can remain empty and an unreserved space can still be occupied. An office can therefore have higher attendance alongside quiet floors, congested collaboration areas and unused room types.

This distinction explains why technology adoption and decision-ready visibility are separate benchmarks. Access systems, reservations and building-wide headcounts can each contribute evidence, but none independently establishes how intensely individual resources are used throughout the day.

The consequences are material for portfolio planning. If a facilities team cannot separate peak crowding from persistent capacity pressure, a shortage concentrated on one weekday or in one room type may look like a need for more total floor area.

Sensors, analytics and AI close successive gaps

Sensors become important when access and booking systems cannot distinguish presence from actual use. People counters can measure movement through a floor, workstation sensors can identify occupied desks, and environmental sensors can add conditions such as temperature or air quality. Each adds evidence, but no single source produces a complete utilization model.

Analytics is the integration layer rather than a substitute for measurement. Comparing locations and periods requires consistent room names, time intervals, capacity definitions and treatment of unavailable areas across access, reservation, sensor and building-management information. A dashboard built on mismatched definitions merely presents the underlying blind spots more neatly.

AI sits downstream of that work. It can identify recurring patterns, automate reports and flag unusual demand, but missing coverage and incompatible inputs limit the reliability of its output. The Johnson Controls OpenBlue facilities report identifies data quality, system integration and cybersecurity concerns as barriers to broader AI expansion.

The resulting sequence is practical: attendance supplies a schedule baseline; occupancy establishes how many people are present; utilization identifies the resources they consume; integrated analytics makes sites and periods comparable; AI can then help interpret those patterns. Skipping a layer converts an unknown into an apparently precise assumption.

Expansion requires measured capacity pressure, not attendance alone

The benchmarks support an operational decision tree rather than a direct jump from more office days to more real estate:

1. If attendance is still changing materially between planning periods, portfolio demand remains unsettled.
2. If attendance is stable but occupancy coverage is incomplete, capacity conclusions remain premature.
3. If occupancy is visible but booking, access and sensor information disagree, definitions and integrations need reconciliation.
4. If reliable utilization data shows pressure only on particular days or in certain space types, scheduling changes or reconfiguration may address the constraint without adding floor area.
5. If comparable measurements show that peak demand repeatedly exceeds usable capacity after those options are assessed, additional space has an evidence-based justification.

The same logic can support a smaller footprint when comparable periods show persistent underuse. Localized shortages may instead indicate that the workplace has the wrong mix of meeting rooms, quiet areas and workstations rather than too little total space.

The August findings do not provide a common measured utilization rate for the workplaces behind the attendance benchmark, and they do not establish that increased attendance improved building efficiency. The missing benchmark is a direct connection between schedules, peak occupancy, duration of use, sensor coverage and demand by space type. Until that evidence exists, five-day attendance is a stronger planning signal—but not proof that an office portfolio is right-sized.

Also read:

• Why Meta’s New Ray-Ban Display AI Glasses Are Still Missing From Europe: The Ironic Cost of “Consumer-Friendly” Regulations
• Five Business Technology Bets for 2026—AI Agents Are Still the Weak Link
• Post Office’s 7.5% Tops SBI’s 6.05% at Five Years