A COO at a 300-person services firm told us her hybrid policy was simple: three days in the office. We asked what happens when someone comes in two. She paused, then said, "Their manager has a conversation with them." We asked what happens in that conversation. She didn't know. Neither did the managers. Six people at that company gave us six different answers, ranging from "nothing" to "it goes in their review."

That gap — between a policy everyone can recite and a practice nobody can describe — is where most flexible work programs actually live. The experiment phase is over. The improvisation phase, unfortunately, is not.

The short answer: Flexibility fails when it is treated as a benefit to be granted rather than an operating model to be designed. The companies that make it work do not have better policies. They have explicit coordination rules, written handoffs, and hard limits on workload — and they enforce those things far more seriously than they enforce attendance. If you want flexibility without burnout, stop legislating where people sit and start engineering how work moves between them.

Flexibility Became a Perk, Not an Operating Model

When offices emptied out in 2020, remote work was a continuity measure. Nobody designed it. It was bolted onto an operating system built entirely around physical co-presence — the hallway question, the whiteboard, the manager who could see who looked overloaded. Those informal mechanisms were doing an enormous amount of invisible coordination work, and when they disappeared, most companies replaced them with meetings and messaging apps.

Six years later, the bolt-on is still a bolt-on. The average mid-market company has a stated policy about days in the office and almost nothing written down about how decisions get made, how work gets handed off across time zones and schedules, or how anyone knows when a team is over capacity. Flexibility got framed as an HR benefit — something the company gives employees — which means it lives in the handbook rather than in the way work actually runs.

The result is predictable. People are technically free to work from anywhere, and functionally unable to stop working anywhere. Meeting load has climbed to fill the coordination vacuum. Managers who can no longer read the room have started reading calendars and green status dots instead. And the people who are best at absorbing ambiguity — the ones who quietly pick up the dropped handoff at 9pm — are the ones burning out first. That is not a flexibility problem. That is an operations problem wearing a flexibility costume.

A Four-Part Operating Model for Flexible Work

The fix is not another policy memo. It is four design decisions, made explicitly, written down, and enforced by managers who have been told what enforcement actually means.

1. Define the unit of coordination, not the number of days

"Three days a week" is a headcount metric masquerading as a coordination rule. It tells you nothing about whether the right people are in the same place at the same time. Replace it with a coordination unit: the team, the project, or the customer account — whichever is the real dependency. Then let each unit publish its own overlap window and in-person cadence based on the work it does.

An implementation team running a live migration may need four hours of daily overlap and one full day together per sprint. A finance team closing the books needs heavy overlap for five days a month and almost none the rest of the time. A design team may need one intense in-person block per quarter and nothing else. Setting a single company-wide number guarantees you are over-constraining most teams and under-serving the ones that actually need to be together. Push the decision down to the unit that owns the dependency, and hold that unit accountable for publishing its rule.

2. Make handoffs explicit and written

In a co-located org, handoffs are ambient. Someone leans over and says "I'm done, it's yours." In a flexible org, that signal has to be manufactured, and if you do not manufacture it, people invent substitutes — usually a meeting, or a Slack message that gets buried, or a follow-up ping at 10pm.

Pick your highest-volume recurring handoffs — the five or six that carry most of your operational load — and write down four things for each: what "done" means, where the artifact lives, who owns it next, and what the response window is. Not a process document. A single card, visible to everyone in the chain. Most teams find they can cover eighty percent of their coordination pain with fewer than ten of these.

The response window matters more than people expect. When it is unstated, everyone defaults to "as fast as possible," which is what actually drives the always-on behavior. A written "next business day" is not a lower standard. It is a standard.

3. Set load limits before you set location rules

Here is the uncomfortable part. Most burnout attributed to remote work is not caused by location at all. It is caused by systematic over-commitment that co-location used to make visible and flexibility now hides. If a team is carrying nine concurrent priorities with capacity for five, that team will burn out in the office, at home, and on the moon.

Before you touch the schedule, count the work. How many active initiatives does each team own? How many were added in the last quarter without anything being removed? What is the actual utilization of your most constrained roles — not the planned number, the observed one? Set a hard ceiling on concurrent priorities per team and make adding one require removing one. This is the single highest-leverage intervention available to you, and it is the one most companies skip because it requires telling an executive no.

4. Measure output and recovery, not presence

Badge swipes, activity tracking, and status-light monitoring do not measure productivity. They measure compliance theater, and every capable employee learns to produce it within about three weeks. Worse, they signal to your best people that you do not trust them, which is a reliable way to lose them.

Track two things instead. First, throughput at the team level: cycle time on your core deliverable, on-time delivery against commitments, quality escapes. Second — and this is the one almost nobody instruments — recovery. Are people taking their time off? Is after-hours activity trending up? Is meeting load per person climbing quarter over quarter? Those three signals will tell you a team is heading for trouble two months before the resignation shows up, and they cost you nothing but a dashboard.

What Founders and CEOs Get Wrong About This

The most common mistake is treating this as a culture question. Leaders describe flexibility debates in terms of trust, commitment, and "whether people want to be here," and those framings feel important but produce nothing actionable. You cannot write a handoff card out of a values statement.

The second mistake is announcing a policy without changing anything underneath it. A mandate to return three days a week, dropped onto an org with no coordination rules and no load limits, produces the same chaos with a longer commute. People come in and sit on video calls with colleagues in other buildings. Nothing improves, trust erodes, and leadership concludes that flexibility was the problem.

The third mistake is confusing autonomy with absence of structure. Founders who value flexibility often resist writing things down because it feels bureaucratic. But structure is what makes autonomy survivable. When the rules are explicit, people can make decisions without checking in. When the rules are implicit, everybody checks in constantly — which is exactly the meeting load they were trying to avoid.

The fourth is assuming this is a one-time project. Coordination requirements change when you add a team, land a large account, or open a second time zone. The companies that keep flexibility working revisit their overlap windows and handoff cards on a regular cadence, usually quarterly, and treat it as ordinary operational maintenance rather than a crisis response.

The Bottom Line

The workplace experiment is finished, and the verdict is not that remote won or the office won. It is that co-location was quietly doing a job — coordination — and any company that removed it without replacing it has been paying for that gap ever since, in meeting load, in dropped handoffs, and in the people who absorb the difference until they leave.

Flexibility is not a benefit you grant. It is an operating model you design, and it costs real work up front: naming your coordination units, writing your handoff rules, capping your concurrent priorities, and instrumenting recovery alongside output. That work is unglamorous and it is not expensive. Most mid-market teams can get through the first pass in a quarter.

What it requires is a decision to stop debating attendance and start fixing the system underneath it. The org that does this does not just avoid burnout. It gets a coordination model that is more resilient than the one it had in 2019, because it is written down instead of living in the hallway.

Frequently Asked Questions

How do we set overlap windows across multiple time zones?

Start from the handoff, not the calendar. Identify which handoffs genuinely require synchronous conversation — usually escalations, ambiguous decisions, and creative problem-solving — and size the overlap to cover those. Everything else should move to written asynchronous handoffs with a stated response window. Most distributed teams discover they need two to three hours of true overlap, not the six they had been forcing.

What if leadership insists on a company-wide in-office mandate?

Do the coordination and load work anyway. A mandate does not fix handoffs or over-commitment, and if you implement the operating model first, you will have data showing which teams actually benefit from in-person time. That is a far stronger position than arguing about the number of days.

How do we spot burnout risk before someone resigns?

Watch three trailing indicators at the team level: unused time off, after-hours activity trend, and meeting hours per person. Any one of them moving in the wrong direction for two consecutive months warrants a conversation with the manager about load, not a conversation with the individual about resilience.

Is it worth doing this if we are under 100 people?

Yes, and it is easier. Small orgs can define coordination units and write handoff cards in a couple of weeks. The companies that struggle most are the ones that scaled past 200 people without ever making these decisions, and then had to unpick years of accumulated informal practice.

If your flexibility policy reads clearly on paper but nobody can describe how the work actually moves, that is a solvable operations problem — and a good place to start. If you would like a second set of eyes on your coordination model, get in touch.