Executive summary
The strongest version of the pro-RTO case is not actually “employees are not producing the work.” In the recent mandates reviewed here, the stated rationales are overwhelmingly about culture, mentorship, learning, collaboration, speed, innovation, and leadership presence. Put differently: the public case for RTO is usually about management wanting more access to the process of work, not just the finished artifact. That does not make the rationale fake. It does mean the rationale is about governance of the employee “black box,” rather than a narrowly measured failure of deliverables such as decks, spreadsheets, analyses, code, or campaign drafts. [1]
The labor market, meanwhile, has not “snapped back” to 2019. Gallup’s remote-capable-job series shows a stable hybrid equilibrium rather than a collapse of flexibility, while BLS time-use data show work-at-home levels still far above pre-pandemic norms. WFH Research’s March 2026 update likewise shows work from home persisting at about 27% of paid full workdays. High-profile RTO mandates are therefore important, but they are not the same thing as a full marketwide reversion. [2]
The productivity literature is also more nuanced than either side’s slogans. The best summary is: fully remote can reduce productivity in some settings, hybrid often does not, and aggregate productivity effects appear modest and context-dependent. A randomized controlled trial at Trip.com found no damage to performance or promotions from two days of home working, but a one-third drop in quits; Stanford/WFH Research estimate fully remote work is associated with about a 10% productivity penalty relative to fully in-person work on average; and the San Francisco Fed finds telework by itself is unlikely to explain major aggregate productivity swings. [3]
The costs of stricter mandates are not evenly distributed. Recent evidence links RTO mandates to higher turnover, slower hiring, and disproportionate exits among women, senior workers, and highly skilled workers. That matters because it means RTO can preserve managerial access while simultaneously degrading the very human capital firms say they want to cultivate. [4]
The most defensible thesis, then, is not “offices never matter.” It is: for digitally legible knowledge work, RTO is usually less a discovery that output requires office presence and more a managerial choice about how closely the employer wants to shape, observe, socialize, and discipline the production process. [5]
The core argument
The “black box” metaphor works best for knowledge work whose value is legible in artifacts: a financial model, a slide deck, a code commit, a product spec, a legal draft, a marketing plan, an underwriting memo, an investment analysis. In those cases, an employer can often observe the input it wants and the output it receives without continuously observing the interior mechanics of how the employee got there. The output is the file, the decision, the recommendation, the design, the release.
That is exactly why recent RTO language is revealing. In the cases below, companies do not usually say, “our files got worse at home,” “our analyses became inaccurate,” or “our deliverables no longer arrived.” They say it is easier in person to strengthen culture, mentor newer staff, learn from one another, brainstorm, invent, move faster, and keep teams connected. Those are all real managerial concerns. But they are concerns about intervening inside the box: coaching, socialization, oversight, synchronization, and norm transmission. [6]
That distinction matters. If the firm’s priority were only the output, then service-level agreements, deadlines, quality thresholds, customer outcomes, and compensation tied to deliverables would do most of the work. But the public rhetoric of RTO is not framed that way. It is framed around proximity, presence, and process. Even Gallup’s own hybrid-work research makes the point indirectly: it stresses that office days are most valuable for complex team tasks, brainstorming, relationship building, mentoring, and feedback, and also finds that manager effectiveness matters far more than location itself. [7]
The thesis should therefore be stated carefully. The claim is not that RTO has nothing to do with performance. It is that, for remote-capable white-collar work, the modern RTO wave is better understood as a governance decision about the labor process than as a simple response to missing deliverables. That interpretation is directly consistent with Ding and Ma’s finding that RTO mandates are associated with managers seeking to reassert control, while failing to produce clear gains in firm value or financial performance. [8]
The policy wave since 2024
The recent RTO wave is best understood not as one homogeneous movement but as a series of employer decisions clustering around similar managerial themes: culture transmission, apprenticeship, collaboration, speed, and co-presence. The visual below shows the major cases covered in this report. Dates are shown at announcement level, with implementation timing where publicly stated. [9]
Major RTO Announcements in the Sample
The recent return-to-office wave did not arrive as one synchronized event. It unfolded as a sequence of large employers tightening expectations around physical presence, usually in the language of culture, speed, learning, and collaboration.
Amazon announces a five-day return
Amazon says corporate employees will return to the office “the way we were before the onset of COVID,” with the five-day expectation beginning January 2, 2025.
Dell tightens sales-team expectations
Dell directs office-capable sales employees toward a five-day office expectation, effective September 30, 2024, while field representatives remain with customers, partners, or in office.
JPMorgan Chase moves most hybrid employees back
JPMorgan Chase tells most employees on hybrid schedules to return to the office five days a week, with the rollout beginning in March 2025.
Starbucks shifts support partners from three to four days
Starbucks tells support-center partners to spend four days a week in office and requires more leaders to relocate to Seattle or Toronto.
Fidelity announces a five-day office move
Fidelity announces that Boston and three other U.S. sites will move to five days in office starting in September 2026.
Comparative table of major recent announcements:
Major Recent Return-to-Office Announcements
The pattern is more revealing than the individual policies. The stated reasons are rarely about failed deliverables. They are about culture, learning, mentorship, collaboration, speed, and managerial access to the process of work.
Fidelity Investments
Affected: 6,200 Boston employees disclosed; three additional U.S. sites also affected. Local coverage reported a broader total around 21,200.
Scope: Boston; Merrimack, New Hampshire; Kentucky; New Mexico.
Policy change: From hybrid to five days in office, with some customer-support phone roles exempt.
Stated rationale: Connection, mentorship, learning, culture, collaboration, and customer delivery.
Amazon
Affected: Corporate employees globally; exact affected count not publicly disclosed.
Scope: Global corporate offices, with local and approved remote exceptions.
Policy change: From at least three days per week to an expectation of office presence outside extenuating circumstances.
Stated rationale: Culture, collaboration, brainstorming, teaching, learning, connection, and speed.
Dell Technologies
Affected: Global sales-team employees able to work from offices; exact affected count not publicly disclosed.
Scope: Global sales organization.
Policy change: Office-capable sales employees to be in office five days; field reps with customers, partners, or in office.
Stated rationale: Collaborative environment, skill growth, and remote work as the exception.
JPMorgan Chase
Affected: Most employees on hybrid schedules; exact affected count not publicly disclosed.
Scope: Firmwide / global.
Policy change: Most hybrid staff directed back to the office five days a week.
Stated rationale: In-office learning, innovation, culture, and solidifying the full-time in-office approach.
Starbucks
Affected: Support-center partners in Seattle and Toronto plus North America regional offices; exact count not publicly disclosed.
Scope: Seattle, Toronto, and North America regional offices.
Policy change: Support partners move from three to four days in office; some people leaders expected to relocate.
Stated rationale: Better idea-sharing, creative problem-solving, speed, and culture.
The pattern is hard to miss. These companies are not mainly saying that remote employees failed to produce files, analyses, decks, code, or recommendations. They are saying that leadership wants more access to the conditions under which those outputs are produced: culture, apprenticeship, correction, synchronization, and managerial presence.
The pattern is hard to miss. In these public statements, the vocabulary is not output failure. It is culture, connection, learning, speed, invention, apprenticeship, and oversight. That is the language of access to the interior of work, not merely the receipt of end products. Amazon’s memo is the cleanest example: it explicitly links office presence to teaching, learning, practice, and culture, not to an empirical claim that deliverables from home had become unacceptable. Fidelity and Starbucks say essentially the same thing in different wording. [19]
This does not prove cynical intent. It does show that, in the most visible mandates, employers are usually making a choice about how work should be organized and supervised, not just what work should be turned in.
What labor-market data say
The broad labor-market picture is more stable than the headlines suggest. Among remote-capable U.S. employees, Gallup found 50% hybrid, 30% exclusively remote, and 20% exclusively on-site in mid-2022; 50% hybrid, 29% remote, and 20% on-site in 2023; 55% hybrid, 26% remote, and 19% on-site in early 2025; and its latest indicator reading available on May 5, 2026 shows 52% hybrid, 26% remote, and 22% on-site. Gallup also says remote-capable jobs make up about half of the U.S. workforce. [20]
BLS data tell a similar story on a broader denominator. The share of employed people who spent some or all of their workday working at home was 34% in 2022, 35% in 2023, and 33% in 2024. That is down from the emergency peak, but still far above pre-pandemic norms. Meanwhile, WFH Research’s April 2026 update estimates that about 27% of paid full workdays in March 2026 were work-from-home days; over the April 2025-March 2026 period, it classifies full-time wage-and-salary employees as 12% fully remote, 26% hybrid, and 62% fully on-site. [21]
The chart below uses Gallup’s remote-capable-job series because it cleanly separates hybrid from fully remote work. Gallup reports that the distribution held broadly stable through 2024; the chart therefore uses its 2022 point, 2023 annual point, 2025 annual point, and the latest indicator reading entering 2026. [20]
What follows from this is important. The market did not discover in 2024-2026 that remote work was impossible. What it discovered in 2020-2022 was that a very large share of white-collar labor could be performed remotely or hybridized. What happened afterward was a negotiation over how much discretion should remain with workers versus employers. Labor-market statistics show that hybrid is sticky. RTO announcements show that some employers want to reclaim more discretion over where and how the work is done. Those are different propositions. [22]
What the productivity literature actually says
The best evidence does not support a crude binary view.
Stanford/WFH Research summarize the landscape this way: working from home rose five-fold from 2019 to 2023; about 40% of U.S. employees were working remotely at least one day a week by then; fully remote work is associated with about a 10% productivity shortfall relative to fully in-person work on average; and hybrid work appears to have no material productivity effect, while helping recruiting and retention. Their updated 2026 paper argues that WFH has stabilized at about 25% of paid workdays, and that forcing the economy back to 2019 norms would lower aggregate labor productivity by 0.2% to 1.1% and worker welfare by about 2.3%. [23]
The most rigorous firm-level evidence for hybrid work is the 2024 Nature paper on Trip.com[24]. In a six-month randomized controlled trial involving 1,612 university-graduate employees in engineering, marketing, accounting, and finance, two days of home working per week improved job satisfaction, cut quit rates by one-third, and did not damage performance reviews, promotions, or lines of code written. The gains were especially strong for women, non-managers, and workers with long commutes. Managers’ beliefs also shifted materially: before the experiment they expected a negative productivity effect; after the experiment they viewed hybrid work slightly positively. [25]
At the macro or sector level, the picture is more muted. The San Francisco Fed concludes that, after controlling for pre-pandemic sector trends, there is little statistical relationship between telework intensity and industry productivity performance during the pandemic era, implying telework by itself is unlikely to explain big changes in aggregate productivity growth. BLS’s productivity office, however, points to recent work finding a positive relationship between total factor productivity and remote work. Those are not contradictory so much as differently scoped: cross-sector correlations can look positive while the aggregate effect remains modest and heavily dependent on task mix, measurement, and management quality. [26]
That is the nuance the headlines usually miss. The serious research does not say “the office is obsolete.” It also does not say “remote work was a mistake.” It says the answer depends on the mode of remote work, the task structure, the social density of the job, the quality of management, the maturity of the employee, and how much of the work product is already legible in digital outputs. That last factor is where the black-box thesis is strongest.
Who bears the cost and what law changes
The strongest empirical challenge to blanket RTO is not a philosophical one. It is a labor-market one. A November 2024 working paper tracking over 3 million tech and finance workers reports that firms adopting RTO mandates saw turnover rise by about 14% on average after the mandate, with more pronounced effects among women, more senior employees, and more skilled employees. Using more than 2 million job postings, the same study finds post-mandate vacancy-filling times increased by about 23% and hire rates fell by about 17%. A separate 2024 paper using resume data on Microsoft, SpaceX, and Apple finds RTO policies can shift firms toward lower tenure and lower seniority, with senior employees exiting to large direct competitors. [4]
That is a meaningful asymmetry. If RTO is justified partly as a way to rebuild mentorship, culture, and institutional knowledge, but it disproportionately pushes out senior and skilled workers, then the policy can erode the very tacit capital it claims to protect. Gallup’s workplace research points in the same direction at the level of sentiment: remote-capable workers forced back on-site showed one of the sharpest drops in feeling respected, falling from 46% to 35% in 2022. [27]
The legal and regulatory implications are also more substantial than casual discussion suggests. Under EEOC guidance, telework can be a reasonable accommodation for disability, but it is not automatically required; if in-person presence is genuinely essential, the employer can insist on it. Under the Fair Labor Standards Act, ordinary home-to-work commuting time is generally not compensable. But remote-work arrangements can trigger employer reimbursement obligations in some jurisdictions: for example, California Labor Code section 2802 requires employers to indemnify employees for necessary expenditures incurred in carrying out their duties. In unionized settings, employers have a duty to bargain in good faith over wages, hours, and other working conditions, and employees—unionized or not—retain NLRA protections to act together over working conditions. [28]
Compensation is the hidden layer here. The San Francisco Fed’s 2026 work on the WFH wage premium finds that jobs involving work from home carry a wage premium, though much of it reflects worker selection rather than a pure causal effect of location. The practical implication is still clear: remote flexibility has economic value. Removing it often changes the employee’s effective compensation bundle even when nominal salary is unchanged, because it changes commute costs, time use, risk of attrition, and the outside-option set in hiring markets. That helps explain why workers frequently experience RTO as more than a scheduling tweak. [29]
AI, the black box, and the strongest counterarguments
AI sharpens the underlying argument in two ways.
First, generative AI makes many white-collar outputs even more legible as units of value. Drafts, summaries, code, spreadsheets, analyses, and design iterations are increasingly digital, versioned, timestamped, and comparable. In principle, that should strengthen output-based management. Gallup reports that AI usage is still uneven—only 15% of white-collar employees say they use AI weekly, though 45% of those users say it makes them more productive and efficient—but the direction is clear: more knowledge work is becoming modular, inspectable, and measurable at the artifact level. [27]
Second, employers are not responding to that increased output legibility by uniformly loosening spatial control. Amazon’s September 2024 memo is revealing because it celebrates growth in GenAI and cost-structure improvement while simultaneously tightening office expectations. The inference is not that AI and RTO are contradictory. It is that, during workflow redesign, firms may want even more access to people for coordination, governance, training, and knowledge transfer—even if the measurable output is increasingly digital. AI can therefore magnify the tension between an output-centric model of value and a management-centric model of supervision. [30]
The strongest counterarguments deserve to be taken seriously. Fully remote work is not neutral in all settings: Stanford/WFH Research estimate a productivity drag for fully remote arrangements on average, and Gallup notes that some activities—complex team tasks, brainstorming, relationship building, mentoring, and feedback—are often easier in person. The Nature/Trip.com study itself excluded interns and employees in probationary periods because on-site learning and mentoring were judged especially important for those groups. In other words, the best evidence does support a real apprenticeship and coordination case for some office presence, especially for early-career employees and highly interdependent work. [31]
That is why the sharpest formulation is not “RTO is just control.” It is more precise than that: RTO is usually a managerial preference for access to the labor process, justified partly by real concerns about apprenticeship, coordination, and cultural reproduction, but only weakly supported by evidence that blanket office presence is necessary for deliverable quality across remote-capable knowledge work. [32]
Recommended framing and headline options
A strong article should avoid saying that office presence never matters. That overstates the case and makes the counterargument too easy. The better framing is:
What 2020 proved was not that offices are useless. It proved that a vast amount of white-collar output can be produced outside the office. What today’s RTO wave shows is that many employers still want something more than the output: they want access to the person producing it, the norms shaping it, and the process through which it is made.
In that framing, “control” is not just surveillance. It includes culture injection, apprenticeship, socialization, live correction, managerial presence, and symbolic authority. That is both more nuanced and more defensible.
Possible headlines:
- RTO Is Less About the File Than the Firm
- If the Deliverable Is Digital, Why Does the Office Matter So Much
- Return to Office and the Fight Over the Employee Black Box
- Remote Work Proved the Output Could Travel; RTO Is About What Management Wants Nearby
- Culture, Control, and the Real Logic of Return to Office
Open questions, limitations, and selected URLs
A few limits matter.
Some companies do not publicly disclose the precise number of employees affected by a mandate. That is why several rows above say “exact count not publicly disclosed.” In Fidelity’s case, the Boston site count of 6,200 is explicitly stated in the source reviewed, while the broader ~21,200 figure appears in local press coverage rather than in a company breakdown I could independently verify from an official Fidelity posting. [33]
The research base is strongest for remote-capable, college-educated, digitizable work. The black-box argument is therefore most persuasive for analysts, designers, coders, marketers, finance staff, product teams, and similar roles. It is much less complete for physical, customer-facing, safety-critical, or place-bound work, where presence is more likely to be part of the product itself or part of regulatory compliance. [34]
The latest public annual labor-market readings also lag the calendar. As of May 5, 2026, the cleanest cross-year Gallup annual point is 2025, while the “2026” figure above is Gallup’s latest indicator reading entering 2026 rather than a full-year annual estimate. BLS annual time-use data currently run through 2024, while WFH Research provides the freshest 2026 monthly pulse. [35]
Selected source URLs:
https://www.aboutamazon.com/news/company-news/ceo-andy-jassy-latest-update-on-amazon-return-to-office-manager-team-ratio
https://www.sec.gov/Archives/edgar/data/1018724/000101872425000004/amzn-20241231.htm
https://www.reuters.com/business/finance/jpmorgan-asks-staff-return-office-five-days-week-march-memo-shows-2025-01-10/
https://www.jpmorganchase.com/content/dam/jpmc/jpmorgan-chase-and-co/investor-relations/documents/quarterly-earnings/2024/4th-quarter/corp-10k-2024.pdf
https://www.reuters.com/technology/dell-asks-global-sales-team-work-five-days-week-office-memo-says-2024-09-26/
https://www.sec.gov/Archives/edgar/data/1571996/000157199625000034/dell-20250131.htm
https://about.starbucks.com/press/2025/message-from-brian-reestablishing-an-in-office-culture/
https://www.sec.gov/Archives/edgar/data/829224/000082922424000057/sbux-20240929.htm
https://www.boston.com/community/readers-say/2026/04/30/readers-react-fidelity-five-day-return-to-office-mandate/
https://www.fidelity.com/about-fidelity/our-company/by-the-numbers
https://www.gallup.com/workplace/397751/returning-office-current-preferred-future-state-remote-work.aspx
https://www.gallup.com/workplace/511994/future-office-arrived-hybrid.aspx
https://www.gallup.com/workplace/657629/post-pandemic-workplace-experiment-continues.aspx
https://www.gallup.com/401384/indicator-hybrid-work.aspx
https://www.bls.gov/opub/ted/2024/35-percent-of-employed-people-did-some-or-all-of-their-work-at-home-on-days-they-worked-in-2023.htm
https://www.bls.gov/news.release/pdf/atus.pdf
https://wfhresearch.com/wp-content/uploads/2026/04/WFHResearch_updates_April2026.pdf
https://siepr.stanford.edu/publications/working-paper/evolution-working-home
https://www.nature.com/articles/s41586-024-07500-2
https://stevenjdavis.com/s/Why-Working-From-Home-Will-Stick-5-February-2026.pdf
https://www.frbsf.org/research-and-insights/publications/economic-letter/2024/01/does-working-from-home-boost-productivity-growth/
https://www.bls.gov/productivity/notices/2024/productivity-and-remote-work.htm
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4675401
https://cdn.arstechnica.net/wp-content/uploads/2024/12/5031481.pdf
https://bfi.uchicago.edu/wp-content/uploads/2024/06/BFI_WP_2024-56.pdf
https://www.eeoc.gov/FAQ-federal-sector-telework-accommodations-disabilities
https://www.eeoc.gov/laws/guidance/work-hometelework-reasonable-accommodation
https://www.dol.gov/agencies/whd/direct-care/travel-time
https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=LAB§ionNum=2802.
https://www.nlrb.gov/about-nlrb/rights-we-protect/the-law/employees/concerted-activity
https://www.nlrb.gov/about-nlrb/rights-we-protect/your-rights/employer-union-rights-and-obligations
https://www.frbsf.org/research-and-insights/publications/working-papers/2026/02/the-work-from-home-wage-premium/
[1] [6] [9] [12] [19] [30] https://www.aboutamazon.com/news/company-news/ceo-andy-jassy-latest-update-on-amazon-return-to-office-manager-team-ratio
[2] [15] [20] [34] https://www.gallup.com/workplace/397751/returning-office-current-preferred-future-state-remote-work.aspx
[3] [23] [31] https://siepr.stanford.edu/publications/working-paper/evolution-working-home
[4] https://cdn.arstechnica.net/wp-content/uploads/2024/12/5031481.pdf
[5] [8] [32] https://papers.ssrn.com/sol3/Delivery.cfm/4675401.pdf?abstractid=4675401
[7] [22] https://www.gallup.com/workplace/511994/future-office-arrived-hybrid.aspx
[10] [33] Readers: Fidelity’s 5-day return-to-office policy ‘incredibly disruptive’
[11] [35] https://www.gallup.com/401384/indicator-hybrid-work.aspx
[13] [25] https://www.nature.com/articles/s41586-024-07500-2
[14] [17] https://www.reuters.com/technology/dell-asks-global-sales-team-work-five-days-week-office-memo-says-2024-09-26/
[16] [24] https://www.reuters.com/business/finance/jpmorgan-asks-staff-return-office-five-days-week-march-memo-shows-2025-01-10/
[18] https://about.starbucks.com/press/2025/message-from-brian-reestablishing-an-in-office-culture/
[21] https://www.bls.gov/news.release/archives/atus_06272024.htm
[26] https://www.frbsf.org/research-and-insights/publications/economic-letter/2024/01/does-working-from-home-boost-productivity-growth/
[27] https://www.gallup.com/workplace/657629/post-pandemic-workplace-experiment-continues.aspx
[28] https://www.eeoc.gov/FAQ-federal-sector-telework-accommodations-disabilities
[29] https://www.frbsf.org/wp-content/uploads/wp2026-02.pdf