Return to Office: How to Run an RTO Transition
How to run a return to office transition: choosing the mandate, setting the days, handling ADA and caregiver exceptions, and enforcing it.
Return to Office
The operational version rather than the trend piece: how to decide whether you need a mandate at all, how to choose the days, how to build the exception process before you announce, where the disability and caregiver exposure actually sits, what notice you owe, and how to enforce the thing consistently once somebody tests it
The first return to office announcement I sent went out on a Thursday afternoon with an effective date eleven days later. By Monday I had two resignation conversations, an accommodation request I had no process for, and a manager quietly telling his team that the rule did not really apply to them.
None of that was about the office. All of it was about the order I did things in. The policy document was fine. It was the last thing I needed and the first thing I wrote. Get the order right and most of that Monday never happens.
A return to office transition is a sequencing problem: justify the mandate, choose the days, define the exceptions before anyone can ask for one, check what notice you owe, announce once, then enforce it through a single path. I build the people and records tooling for small businesses without an HR department at FirstHR, and this is the order I would use now.
Treat what follows as general information rather than legal advice. A change of this size is worth twenty minutes with employment counsel in your state.
What RTO Means in Practice
RTO stands for return to office. It means requiring employees who have been working remotely, in whole or in part, to perform their work at a company location again, either every scheduled day or on a defined number of days each week.
That distinction matters more than it sounds. I have seen a single document try to cover both, and the result was an attendance policy carrying leave language it could not support. Keep the office attendance rule separate from anything touching medical leave, and if you want the drafting done for you, the return to office policy templates are already scoped that way.
The second thing to know before you decide anything is where the market actually sits, because an employer that still has remote staff tends to assume it is an outlier. According to the Bureau of Labor Statistics American Time Use Survey (2025 results), 35 percent of employed people did some or all of their work at home on days they worked, which is not the picture the headlines describe.
According to Gallup (May 2026), 52 percent of remote-capable US employees work hybrid, 26 percent work exclusively remote, and 22 percent are fully on site (Gallup). Gallup reports that this split has held relatively steady since 2022.
Return to Office Trends and What They Mean for a Small Employer
Return to office trends point in several directions at once, which is why the coverage reads as contradictory. Fully remote and fully on site each account for roughly a quarter of the remote-capable workforce, and hybrid still holds a lead that none of the recent movement comes close to threatening.
Notice who the five-day headlines belong to. They come from employers with large buildings, long leases, and a workforce big enough that a rule has to be uniform to be administered at all. None of those conditions describe a company without an HR department, and the reasoning that makes a five-day mandate defensible there does not carry over to yours.
The shift I actually see among small employers has nothing to do with the day count. What changes is that the arrangement gets written down at all: a named model, a dated policy, and an exception route with somebody's name attached to it. That change does more for a small team than matching whatever number the latest large-employer headline announced.
So there is no wave here to catch or to miss. Trend data earns its place in one part of this decision: calibrating how harsh your mandate is against what your competition for the same hire is actually running, rather than against a market where everyone has supposedly gone back already. It cannot tell you what your own mandate is for, which is the next question.
Decide What the Mandate Is For
Before choosing days, write one sentence describing the behavior you want that is not happening now. If you cannot write it without using the words culture or collaboration, you do not yet have a reason you can defend to the person who has to rearrange their childcare.
The test I use is whether the sentence contains something observable. "Engineers and support never see the same problem at the same time" is observable. "We have lost our culture" is a feeling that may be true and cannot be scheduled against.
The first example tells you which people need to overlap and on which days. The second tells you nothing, which is why it usually produces a five-day mandate.
The cost side is real. A mandate changes terms people accepted, and the people best placed to walk away over it are senior staff and specialists, because they have the most options. None of that rules out a mandate. It does mean you should know what you are buying, because high turnover lands on the hardest people to replace.
| Stated reason | Is it observable? | What it should drive |
|---|---|---|
| Cross-team problems take days to resolve | Yes, cycle time is measurable | Shared anchor days for the two teams involved, not a company rule |
| New hires ramp slowly and ask nobody for help | Yes, ramp time and early attrition | On-site requirement weighted to the first months rather than forever |
| Customers or equipment are on site | Yes, the work is physically located | Role-by-role designation written into the job description |
| Managers cannot tell who is working | No, this is a management problem | Clear expectations and a review cadence, not a commute |
| We have lost our culture | No | Nothing until it is restated as an observable behavior |
| The lease is expensive and the office is empty | Yes, but it is a real estate decision | An honest conversation about the lease before a mandate |
The last row is the one people are least willing to say out loud, and employees can smell it from a distance. If the driver is a lease you signed before the world changed, saying so is better received than a collaboration argument nobody believes. Honesty about a commercial reason survives contact with a skeptical team. A borrowed justification does not.
Setting the Days
Pick the smallest number of on-site days that produces the overlap you named, and where overlap is the point, name the specific days rather than the count. A minimum-days rule with free choice is the most popular model and the one most likely to fail at its own purpose.
Here is the failure mode nobody predicts. You require two days a week and let teams choose. Engineering picks Tuesday and Wednesday because of a standing planning meeting. Support picks Wednesday and Thursday because of shift coverage.
The two teams that most needed to see each other now overlap for one day, and half of each person's office days produce none of the overlap the rule was for. The structure of the hybrid schedule decides whether the mandate works, more than the number does.
Midweek days are the compromise most employers land on, and there is a reason. Monday and Friday attract the highest exception traffic, carry the most childcare conflict, and generate the most resentment per commute. If your goal is overlap rather than symbolism, a pair of midweek days will get you further than a Monday requirement chosen to make a point.
Also decide whether one rule should cover everyone. A company-wide standard is simpler to explain and harder to defend when a role genuinely does not benefit from it. Role-by-role designation is more defensible, but it means writing down a reason for each designation, which is work most small employers skip and then regret. Whichever you pick, it belongs in your written company policy set rather than in a manager's head.
Build the Exception Process First
Decide who may request an exception, who decides, what evidence is required, how long a decision takes, and whether an approval is permanent or reviewed, all before the announcement goes out. An exception process invented in response to the first request will look like favoritism no matter how fair it actually is.
The reason is social, not legal. The first exception you grant becomes the precedent everyone else measures their own case against. If it was granted informally by one manager over a coffee, that is the standard you have set, and you will spend the next quarter explaining why somebody else with a similar story got a different answer.
The response time item is the one small employers underestimate. During my own transition the single largest source of goodwill was answering exception requests within a week, including the ones I declined. A fast no is easier to absorb than a slow maybe, and it lets people plan.
Give the decision owner one form to fill in for every request, so that six months later you can show what was asked, which criteria were applied, and why the answer was what it was. Fill in your own criteria before the announcement and reuse the same copy for every case.
Where the Legal Exposure Sits
The mandate itself is rarely the legal problem. In most states employment is at will, meaning either side can end it for any lawful reason, and work location is a term the employer sets. The exposure comes from how individual requests are handled, and it concentrates in four places.
On disability, the framework is well established. According to the Equal Employment Opportunity Commission, the Americans with Disabilities Act does not require an employer to offer telework to everyone. Where an employer does offer it, employees with disabilities must have an equal opportunity to participate. Allowing an employee to work at home may also be a reasonable accommodation when a disability prevents performing the job on site and the work can be done at home without significant difficulty or expense (EEOC).
Two practical consequences follow. First, if you already allowed remote work broadly, it becomes harder to argue that on-site presence is an essential function of a specific job, meaning one of the fundamental duties of the position. Second, you may explore effective on-site accommodations before agreeing to remote work.
The Commission's enforcement guidance on reasonable accommodation is explicit about that second point. The employee's preference gets primary consideration, but where two accommodations would both be effective, the employer has the ultimate discretion to choose between them (EEOC). What you cannot do is decline without assessing the specific job.
Pregnancy sits in its own lane. According to the Equal Employment Opportunity Commission, the Pregnant Workers Fairness Act requires employers with 15 or more employees to reasonably accommodate known limitations related to pregnancy, childbirth, and related medical conditions (EEOC).
Telework and changes to a work schedule both appear on the EEOC's list of example accommodations under the Act, and a return to office change is exactly the event that surfaces requests for them. Read the Pregnant Workers Fairness Act requirements before you set an effective date rather than while you are answering the first request.
Caregiver exposure works differently and catches more small employers. Caregiver status is not itself protected under federal law. The Commission has nonetheless warned for years that unfavorable treatment of workers with caregiving responsibilities can constitute unlawful disparate treatment where it is based on a protected characteristic such as sex. That includes decisions driven by assumptions about which parent wants flexibility (EEOC enforcement guidance). The safe pattern is written criteria, applied identically, with the reasoning on file.
The Family and Medical Leave Act raises a separate question. An employee using intermittent leave, meaning time off taken in separate blocks rather than all at once, has protections that no attendance policy overrides, which is worth checking against your FMLA obligations if you are covered.
What Notice You Owe
No general federal rule requires advance notice before changing where employees work. Several state and local rules apply around the edges, and your own documents may create obligations that statutes do not.
| Source of obligation | What it can require | Who it typically reaches |
|---|---|---|
| State wage notice laws | Written notice before a change to the information in the wage notice takes effect, in New York at least seven calendar days unless the change appears on the wage statement | All employers in the state, triggered by changes to the rate or basis of pay |
| Predictive scheduling laws | A written schedule a set number of days in advance plus premium pay when it changes late | Large employers in covered hourly industries such as retail, hospitality, and food service in Oregon and several cities |
| Employment agreements and offer letters | Whatever the document promised about location or notice | Individual employees, and often the most senior ones |
| Handbook and policy language | Notice periods your own handbook committed to | Everyone who received the handbook |
| Collective bargaining agreements | Bargaining obligations before a unilateral change | Represented workforces |
| Practical runway | Nothing legally, but transitions announced with less than a month land badly | Everyone, every time |
Oregon is the state to know on scheduling. Its predictive scheduling law requires covered employers to provide written work schedules at least 14 calendar days in advance and to pay an extra hour at the regular rate when the employer adds more than 30 minutes of work to a shift (Oregon Bureau of Labor and Industries).
Coverage there is narrow. The law reaches retail, hospitality, and food service employers with at least 500 employees worldwide, so it rarely bites an office return, though it can if you run mixed operations.
The pay-notice trap catches office employers more often. If your remote arrangement came with a home internet stipend, an equipment allowance, or a different rate, check whether ending it changes anything you listed in the wage notice, the written statement of pay terms you gave the employee.
New York Labor Law 195 requires written notice at least seven calendar days before a change to that information takes effect, unless the change appears on the wage statement (New York State Senate). Treat any money attached to remote work as a separate change with its own notice, documented alongside the schedule change itself.
Announcing It
Announce once, in writing, to everyone at the same time, with a real effective date and a named route for exceptions. Every hour between the first leak and the official message gets filled with a worse version of the story than the one you are about to tell.
Small employers often skip the acknowledgment step, and it is the one that matters most later. A signed acknowledgment stored in the employee record is the difference between an expectation you can enforce and a memo somebody credibly claims never to have seen. If your employee handbook already has an acknowledgment flow, run this through the same one.
One more thing about tone. The message that landed best for me admitted the trade openly: this costs you something, here is what we are getting for it, here is what we are doing to make the days worth the commute.
Announcements that pretend nobody is losing anything get read as either dishonest or unaware, and both are worse than the actual news. Change of this size is a change management exercise before it is a policy exercise.
What a Return to Office Policy Has to Contain
The document you attach to that message needs six things: who it covers, the model and the named days, the effective date, the exception route with its criteria and response time, what happens when somebody does not comply, and a line for the signed acknowledgment. Everything past that is commentary.
The two most often missing are the exception route and the enforcement path, and they are the two that decide whether the policy survives contact with a real request. A document that names the days but never names the decision owner sends every awkward case to whichever manager is standing nearest.
Keep it short and keep it dated. A one-page policy people actually read beats a five-page one that gets skimmed, and the version date is what tells you two years later which rule was in force when a particular exception was decided.
Enforcing It Without Losing People
Enforce through one documented path with one reviewer rather than through manager discretion. Inconsistent enforcement is simultaneously the biggest legal exposure and the fastest way to lose the people who are complying.
The sequence that works is unremarkable: a private conversation first, then a written reminder, then formal discipline, with the same reviewer signing off at each stage. Its strength is that everyone goes through the same steps.
A team that watches two people skip the office with no consequence while a third gets a written warning has learned that the rule is about who your manager likes.
Before you turn on any attendance tracking, decide what you are measuring and why. Recording who is on site on which day is reasonable and often necessary for space planning. Using it as the primary performance signal is not, and it changes behavior in ways that produce compliance without producing the overlap you wanted. Keep any attendance record proportionate to the reason you started it.
Finally, accept that some attrition is the price and decide in advance which of it you can absorb. The people most able to leave over a commute are the ones with the most options elsewhere, which is rarely the group you budgeted to lose.
Two departures from that group can break a small team. If yours is one of them, the honest choices are a softer mandate, a role-designated exception, or a retention conversation before the announcement rather than after the resignation.
What to Track After the Effective Date
Track four things for the first two quarters: whether the behavior you named actually improved, how many exceptions were requested and granted, whether resignations rose in specific teams, and whether enforcement was consistent across managers.
The first one is the one everybody forgets. You wrote a sentence describing an observable behavior that was not happening. Go back and check whether it is happening now. If cross-team resolution time has not moved after a quarter of shared anchor days, the days are not the constraint, and adding a third day will not fix a problem that was never about location.
Exception volume tells you whether the mandate fit the workforce. A handful of requests suggests the policy matched reality. A flood suggests you set the bar somewhere your people cannot reach, and it is cheaper to adjust the policy than to process a hundred individual negotiations. Pay attention to which teams the requests cluster in, because that pattern usually points at a role that should have been designated differently.
Resignations by team show where the mandate cost more than you planned. Compare them against the attrition you decided in advance you could absorb. If departures cluster in one team, or among the people you least wanted to lose, revisit that team's model or designation before the next quarter.
Consistency is the one that protects you. Once a quarter, read the exception log end to end and ask whether two similar requests got two different answers. If they did, find out why while the answer is still an administrative correction rather than an exhibit in a legal claim. That review takes half an hour and is the single highest-value thing on this list.
Two tabs are enough to run all of that: one row per exception request, and one row per attendance escalation with the reviewer named on every line. Keep both in one file so the quarterly read is a single sitting.
| A | B | C | D | E | F | G | H | I | J | K | L | M | N | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Reference | Date received | Employee | Team | Type (accommodation, hardship, preference) | What was requested | Decision owner | Decision | Reason recorded | Date answered | Days to answer | Permanent or reviewed | Review date | Similar case reference |
| 2 | ||||||||||||||
| 3 | ||||||||||||||
| 4 | ||||||||||||||
| 5 | ||||||||||||||
| 6 | ||||||||||||||
| 7 | ||||||||||||||
| 8 | ||||||||||||||
| 9 | ||||||||||||||
| 10 | ||||||||||||||
| 11 | ||||||||||||||
| 12 | Read this sheet end to end once a quarter and ask whether two similar requests got two different answers |
Where Small Employers Get This Wrong
Five patterns, and the first one is the one I ran into personally.
Announcing before the exception process exists is first. The requests arrive within hours, you improvise the first few answers, and those improvised answers become the precedent for every case after them.
Writing the policy before deciding the reason is second. The document is the easy part, and doing it first lets you skip the only genuinely hard question, which is what behavior you are trying to change.
Letting managers grant local exemptions is third. It feels generous in the moment and it destroys the rule for everyone else, while creating exactly the inconsistent treatment that turns an attendance matter into a discrimination claim.
Treating an accommodation request as a complaint is fourth. Nobody uses the legal words. A manager who answers a health-related objection with a restatement of the policy has just skipped the interactive process on your behalf.
Giving too little runway is fifth. Childcare, school pickup, commuting arrangements, and in some cases housing all take longer than two weeks to rearrange, and a compressed effective date converts people who would have complied into people who resign.
Run that same test before you announce anything: write your exception criteria, then put three realistic requests from your own team through them on paper. Whatever answer makes you uncomfortable is the part of the policy to fix first.
Frequently Asked Questions
What does RTO mean?
RTO is short for return to office: an employer decision that staff who have been working remotely, all of the time or some of it, must go back to doing their jobs at a company site. How much on-site time the rule demands varies. Some employers want people in every working day, some set a weekly day count, and some pick specific days when the whole company is in together. Do not confuse RTO with return to work, the term for an employee resuming duties after medical leave, a workers compensation claim, or another protected absence. People mix the two up all the time. Return to work brings leave and disability obligations that an office attendance rule does not, so keep them in separate policy documents rather than one combined draft.
Can an employer legally require employees to return to the office?
Generally yes. Because employment is at will in most US states, the employer decides where the work happens, and an on-site requirement is lawful in principle. The legal limits are specific rather than general. A union contract or an individual employment agreement may promise remote work. A disability, or a limitation tied to pregnancy or childbirth, may create a duty to consider remote work as a reasonable accommodation. And if similar employees are treated differently along lines that follow a protected characteristic, you have exposure no matter how neutral the written policy looks. In practice, legal trouble seldom comes from the requirement to be on site. It comes from how you handle the individual requests that follow it.
How many days a week should a return to office policy require?
No single number is right, but there is a right way to find yours: require the fewest days that still create the overlap you actually need. When the problem is that colleagues who depend on each other are never in the same room, which days matters more than how many, so name shared anchor days instead of requiring an unnamed count. When the problem is access to customers or equipment, the work itself sets the number. Gallup research on remote-capable US employees shows hybrid remains the most common arrangement rather than a transitional stage, so a policy built around a realistic mid-week overlap tends to hold up better than a maximal one.
Do you have to give employees notice before changing to in-office work?
No, not as a matter of federal law, which sets no general advance-notice requirement for a change in work location. Some state and local rules still apply, though. In New York, when a change touches the information in an employee’s wage notice, such as the rate or basis of pay, the employer must give written notice at least seven calendar days before it takes effect, unless the change shows up on the wage statement. Oregon and a number of cities have predictive scheduling rules that call for written schedules in advance and premium pay for late changes, but those rules reach large employers in hourly industries such as retail, hospitality, and food service rather than office roles. Your own paperwork counts too: a contract, an offer letter, or handbook language can create a notice obligation of its own.
What if an employee asks to stay remote as a disability accommodation?
Treat it as an accommodation request and begin the interactive process, the back-and-forth in which you and the employee work out what would help, even when it arrives as a complaint about the new policy rather than a formal request. The Equal Employment Opportunity Commission’s position is that the Americans with Disabilities Act does not oblige an employer to make telework available to all staff. If you do allow it, though, employees with disabilities need an equal chance to use it, and letting someone work from home can itself be a reasonable accommodation when a disability stops them from doing the job on site. You can look at on-site alternatives first, and you can turn down remote work where attendance is genuinely an essential function, but keep a written record of how you reached that decision.
Can you deny remote work to an employee with childcare responsibilities?
Usually yes. Federal employment discrimination law does not treat caregiver status as a protected class, so the risk lies in how you compare people rather than in any single refusal. For years the Equal Employment Opportunity Commission has cautioned that worse treatment of employees with caregiving duties can be unlawful discrimination if it rests on a protected characteristic such as sex, and that covers guessing which parent will want flexibility. Decide every caregiver request against the same written criteria, record why you decided as you did, and never base the answer on an assumption about what a particular employee probably wants.
How do you enforce a return to office policy without losing people?
Put every attendance problem on one written track instead of leaving it to each manager, and keep the attendance question apart from any wider people issue. The track should escalate identically for everyone: talk privately first, follow up with a written reminder, and move to formal discipline only after that, with one reviewer approving each step. The value comes from the uniformity more than from the steps themselves. When a team sees one person excused and another disciplined for the same absence, it concludes that the rule belongs to the manager rather than to the policy. Inconsistency also hits retention unevenly. The employees who can most easily leave over an attendance dispute are the ones with the most options elsewhere, so uneven enforcement costs you most among the people you least want to replace.