August has produced two kinds of announcement, often from the same organisations in the same week. The first is a postponement: return-to-office dates across the largest technology employers have slid into October, then into January, as the Delta wave rearranged everyone's autumn. The second is a mandate: two days a week, or three, or "most of the week", published as settled policy. Both come from the same source, and it is not analysis. It is a calendar, a lease, and a senior team's intuition about what a company should feel like. This matters because hybrid work policy is being written once and treated as permanent, at the exact moment when the evidence needed to write it well has never been more available and less examined.
The four questions a policy has to answer
Strip away the tone and every hybrid policy answers four questions. What is the office for? Which work genuinely requires people in the same room? How much space do we therefore need? And who decides — the company, the function, the team, or the individual? Read the policies published this month against that list. Most answer the third question implicitly, by reference to a lease that expires in 2026. Most answer the first rhetorically, with the word culture. Very few answer the second at all, which is remarkable, because it is the only one that has a factual answer.
The data almost everyone already has
The striking thing about hybrid policy-making is that organisations commission surveys about preferences while ignoring six datasets already sitting in their own systems. Access records. Aggregate daily occupancy by floor and by team, for this year and for 2019. Most offices were never uniformly full: Tuesday to Thursday peaked and Friday was thin long before anybody said hybrid. If your building ran at seventy per cent on its best day in 2019, a three-day mandate is not a return to anything. Calendar metadata. Meeting volume, average attendee count, the ratio of intra-team to cross-team meetings, and how many meetings include somebody external. This tells you which teams have a coordination problem and which have a meeting habit. Room booking and no-show data. The cheapest available signal about what the office is actually used for, and the fastest way to discover that you need three times as many four-person rooms and half as many boardrooms. Remote access and device location. Aggregate, by team, by day of week — enough to know the real pattern rather than the reported one. Throughput by day. Cases closed, invoices processed, tickets resolved, deals advanced, by day of week over eighteen months. If someone claims productivity collapsed, this is where it would show. Joiner and leaver data by team, with tenure. The people most plausibly harmed by distance are those in their first six months, and that is measurable rather than assertable. None of this requires new tooling, and none of it requires looking at individuals. The unit of analysis is the team and the week, never the person and the day. That distinction is the whole difference between workplace analysis and staff monitoring, and it should be stated in writing before anybody runs a query.
| Evidence | Planning question |
|---|---|
| Occupancy | What peak demand must the space accommodate? |
| Room bookings | Which activities need a room? |
| Calendar patterns | Where do teams need coordination? |
| Throughput | What changes alongside the work pattern? |
| Joiner experience | Where does induction need more support? |
Qualitative summary of this article's source text, not a measured outcome or performance estimate.
Why preference surveys mislead
The standard instrument is a survey asking how many days people would like to be in the office. It reliably produces a number around two and a half, which fits nobody and describes nothing. The problem is well documented this year: people simultaneously want more flexibility and more time with colleagues, and the two desires are not reconcilable in a single ratio. Stated preference also drifts with the news cycle, and it varies enormously by commute, seniority, household and role — so an organisational average is an artefact of your demographics, not a finding. Better questions exist. Ask about behaviour rather than desire: which days were you in the office last month? Ask about activities rather than ratios: where would you rather run a design review, a difficult client call, a first-week induction, four hours of analysis? Ask managers about coordination failures rather than satisfaction: what decision took two weeks that should have taken two days, and where were the people? Those answers point at a design. The ratio question points at a compromise.
Design the week around the work
The useful output is not a number of days. It is a statement about which activities are scheduled for co-presence, decided at team level within a company-level frame. The principle worth adopting is that the office should be scheduled for the work that fails asynchronously: onboarding and apprenticeship, contentious decisions with real disagreement, early-stage creative work with no artefact yet, relationship-building with clients and with new colleagues, and incident response. Individual analytical work, writing, code, reconciliation and most status reporting do not need a room, and requiring them in one produces a commute in exchange for a worse desk. That approach also fixes the coordination failure that undermines most mandates. A policy that says three days without saying which days produces offices full of people on video calls with colleagues who chose different days — the worst of both models, at the cost of both. Anchor days set by team, published in advance, with the company setting only the outer boundaries, get the co-presence without pretending everyone's work is identical.
The arithmetic nobody publishes
Here is the trade-off in plain terms. If every employee must attend three days and is free to choose which, almost everyone will choose Tuesday, Wednesday and Thursday. Peak demand then approaches your full pre-2020 desk count on three days and leaves the building near-empty on two. You have kept the entire property cost and added the friction of hybrid. The alternatives are to stagger days by team, which reduces the peak but also reduces cross-team encounter — the thing the mandate was for — or to accept a smaller estate with a booking system and genuine variation in what the space contains: fewer assigned desks, far more small meeting space, and quiet rooms for the calls people currently take in corridors. Choose deliberately. What organisations are doing instead is choosing the first option accidentally and then complaining about the property bill.
Practical Guidance for Work Policy Design
- Pull access, calendar, room, throughput and joiner data first — aggregate only, team-level, two weeks of work, before any policy drafting.
- Compare 2019 occupancy with your target, because the baseline was never full and the mandate is often more attendance than you had.
- Ask about last month's behaviour and specific activities, not preferred ratios.
- Name the activities that are scheduled for the office, and let teams set anchor days inside a company frame.
- Publish the desk arithmetic so the choice between staggered days and a smaller estate is made consciously.
- Version the policy with an owner and a ninety-day review date, and name the three metrics that would cause you to change it.
- Write down the analytics boundary — team and week, never individual and day — and say so publicly.
- Measure promotion and stretch-assignment rates against attendance after a year, because proximity bias is the most likely hidden cost and the only way to find it is to look.
The Regional Angle
The first thing a Gulf employer should notice is that the global hybrid conversation is written for people with a spare room. A policy that treats home as the default and the office as the concession is a policy for the senior third of the workforce. Across this region a large proportion of employees live in shared accommodation, in company housing, or with extended family in a flat where the dining table is contested at ten in the morning. For them the office is not a cost imposed by the employer; it is air conditioning, a desk, a reliable connection, lunch and a door. Some of the most articulate objections to remote work in regional organisations come from people whose managers assume they are asking for a commute. The consequence for planning is concrete: desk demand here will not fall as far as the global benchmarks suggest, the office reduction case is weaker, and a genuinely fair policy sets a floor of guaranteed workspace for anyone who wants it rather than treating attendance purely as an obligation. The second is that location flexibility collides with the structure of employment here in a way it does not elsewhere. Packages are built around housing and schooling in a specific city, residency is tied to a local employer, and payroll must run through the local entity under the wage protection arrangements. Once you tell people that work is not a place, a predictable set of requests follows: to spend the summer working from the home country, to relocate to a cheaper emirate, to keep the role and leave the region entirely. Each one raises questions about residency validity, payroll obligations, permanent establishment and insurance that the policy document does not mention, and the honest answer is usually that the arrangement is possible for a defined number of weeks and impossible beyond it. The inbound mirror of the same trend is already live: this year's virtual working programme lets foreign employees of foreign companies live here while working remotely, which tells you that governments in the region are thinking about the mobility of work considerably more precisely than most employers are. Put the rule in writing before somebody's spouse signs a lease in another country. The third is that a large share of regional work is calibrated to somebody else's attendance. Government and semi-government counterparties returned to the desk earlier and more completely than the private sector, and much of the commercial life of this market still happens at counters, in majlis meetings and at site. If your function depends on ministry submissions, bank branches, customs, courts or client sites, your anchor days are not a cultural choice — they are set by the people you need to be in front of. Map that dependency by function before publishing a company-wide ratio, or the policy will be broken in week one by the departments that hold the licences.
The objection worth taking seriously
The strongest objection comes from experienced operators: data-driven work policy is a euphemism for delay, and delay is worse than a rule people dislike. Employees want to know where to be on Monday. A steering group, a survey vendor, a pilot and a six-month analysis produce ambiguity, which lands hardest on the people with least power to interpret it. And measuring how work happens has a bad history — it slides toward monitoring faster than anyone intends, and once staff believe attendance is being scored, the policy conversation is over. Both halves of that are right, and the second is the reason to be explicit about the analytics boundary rather than quiet about it. But the objection assumes analysis is slow, and the datasets described above are not. Access logs, calendar counts, room bookings and throughput by weekday can be assembled in a fortnight by people you already employ, because the data has been accumulating since 2019 without anybody reading it. The choice is not between deciding quickly and studying carefully; it is between deciding with the evidence in your own systems and deciding without it. Decide in a week if you like — clarity has real value — but publish the metrics you will review at ninety days, and stop describing a first attempt as a permanent settlement. The failure mode of 2021 will not be policies that were too slow. It will be policies that were announced as final, contradicted by the October wave, and quietly abandoned team by team, leaving employees with a document nobody follows and managers improvising in private.
Common Questions
Is there a right number of office days?
No. There are right answers about which activities need a room, and a desk arithmetic that follows from them. The ratio is an output, not an input.
Can we look at badge and calendar data without it becoming surveillance?
Yes, if the unit of analysis is the team and the week, the outputs are aggregate, no individual report is produced, and you say all of that before you start. Break any of those conditions and you have a different project with a different reputation.
What about people who simply will not come in?
Handle it as a management conversation about specific commitments — named activities, named days — rather than as an enforcement problem about a ratio. And check first whether the activities on those days actually required the room.
What should we expect over the next twelve months?
Expect the January dates to slip again, and expect mandates published this month to be softened function by function without a formal announcement. Expect the desk arithmetic to force coordinated anchor days at most large employers by the middle of next year, because the alternative is paying for an estate that is full on three days. Expect vaccination status and entry requirements, not day counts, to be the practical operational question through the autumn. Expect location flexibility to turn into a compensation and mobility issue as employees ask to be paid in one place and live in another, with the first tax and residency surprises arriving shortly after. And expect the first serious evidence on proximity bias in promotions to appear in 2022, at which point the organisations that recorded attendance and progression together will be able to answer the question, and everybody else will be guessing again.
Work Policy Design Session — we assemble the occupancy, calendar and throughput evidence you already hold, translate it into an activity-based hybrid design, and write the policy with a review date instead of a promise.