Flexibility Is Turning the Workplace into a Real-Time Operating System

Most organisations managing flex office environments believe they have a reasonable picture of how their spaces are used. What they actually have, in most cases, is an estimated picture — built from self-reported attendance, occasional surveys, badge access aggregates and local management perception.

The distinction matters more than it appears.

The AREMIS Hybrid Work Report 2026 reveals a consistent pattern: organisations that estimate their office attendance report an average of 56%. Those that actually measure it report 51%. A five-point gap between two populations — those who count and those who guess — that compounds across every space planning decision, every lease renewal, every desk-sharing ratio calibrated on the wrong baseline.

Declarative data has its own ceiling. In a separate analysis, AREMIS conducted direct physical measurement across four multi-tenant office buildings under its operational mandate — three in Brussels, one in Paris — covering 48 tenants. These buildings and their occupants are not part of the HWR survey panel, so the figures are not directly comparable. But the direction is unambiguous: actual measured occupancy averaged 36%, well below the declarative figures observed in the study.

The executive implication is straightforward and largely unacknowledged: organisations are making portfolio decisions from a systematically overstated baseline. Not because the data is unavailable — but because declarative measurement conflates what people think happens with what actually does.

The Geneva case reframes the problem

In June 2025, following repeated employee complaints about overcrowding and unavailable desks, AREMIS conducted a physical headcount in a Geneva multi-tenant office building operating under a standard flex office model — no assigned desks, dynamic access.

The building had no capacity problem. What it had was a distribution problem: presence concentrated unevenly across zones and days, compounded by weak clean desk practices that made available capacity invisible to arriving employees.

This distinction is not semantic. It changes the entire intervention logic. Adding desks solves nothing. Reducing square meters makes things worse. The lever is operational: redistributing presence, enforcing usage norms, dynamically signalling available space. The issue was never volume. It was the absence of a system capable of orchestrating how space is actually used in real time.

Most organisations facing similar complaints would have responded with a space audit. The Geneva case suggests the more relevant question is: do we have the operational mechanisms to manage a shared environment under variable demand?

Flex office transferred management complexity downward — without transferring tools

The shift from assigned to shared workspaces looks like simplification on paper. Fewer desks. Lower occupancy ratios. Reduced real estate cost. What it actually does is transfer a management problem from the organisation to the employee.

In an assigned-desk model, spatial arbitration happens once — at allocation. In a flex model, it happens continuously, every morning, by every employee, without coordination. The organisation gains nominal efficiency and loses operational control.

The frictions that follow — crowded Tuesdays, empty Fridays, meeting rooms monopolised by individual heads-down work, collaboration zones used as storage — are not behavioural failures. They are the predictable consequence of introducing a dynamic access model without the governance layer to run it.

The organisations struggling most with flex office are rarely those that lack space. They are those that treat shared space as a passive infrastructure problem when it has become an active coordination problem. Mutualisation requires arbitration. Arbitration requires data. Data requires systems that most workplace environments were not designed to produce.

What operational steering actually requires

The instinct when occupancy data reveals problems is to add dashboards. That instinct is usually wrong — or at least insufficient.

Reporting and steering are different capabilities. Reporting describes what happened. Steering changes what happens next. The gap between them is where most workplace data investments currently stall: organisations can describe their occupancy after the fact but lack the closed feedback loops to act on it operationally.

Operational steering requires three things that rarely coexist. Reliable measurement that distinguishes zone saturation from global averages and peak demand from smoothed weekly figures. Contextualisation that connects occupancy signals to team allocation, collaboration patterns and cost structures — so the data reveals not just whether space is occupied but whether it is performing. And crucially, closed feedback loops: the organisational capacity to act on signals continuously rather than quarterly.

That last element is where most organisations are furthest behind. The data infrastructure often exists in fragments — sensors, badge systems, reservation platforms, FM ticketing. What is missing is the integration layer that turns these signals into coherent operational decisions: reconfiguring underperforming zones, smoothing attendance peaks, adjusting service levels before friction becomes visible to employees.

The workplace that feels effortless to employees is the one being actively steered underneath. The technology succeeds precisely when no one notices it.

The operating model question

What the data ultimately reveals is a governance gap rather than a space gap. Organisations have restructured their real estate around flexibility without restructuring the management systems required to operate flexible environments. The result is that workplace demand is now variable, but workplace management remains periodic.

That mismatch has a cost — not always visible in real estate budgets, but accumulating steadily in coordination friction, execution overhead and the quiet erosion of employee experience that precedes disengagement from the office model itself.

The organisations closing this gap are not necessarily those with the most sophisticated sensor infrastructure. They are those that have accepted a more fundamental shift: the workplace is no longer passive infrastructure to be reviewed annually.

It is a dynamic operating layer — and it requires the same management cadence as any other system that directly shapes how work gets done.