Office buildings in the 10 major US cities tracked by keycard data ran at 54.7% of their early-2020 occupancy in the last full week of September 2026, even though one in three US firms now requires full-time office presence (Kastle Systems, Back to Work Barometer 2026; Flex Index 2025). The gap between policy and behavior is the story of hybrid work in 2026: required office time has climbed 12% since early 2024 while actual attendance rose only 1-3%, according to the Flex Index. Midweek still carries the load, with Tuesday peaking at 65.3% in Kastle’s data, and 43% of hybrid workers admit to “coffee badging” visits that register a swipe without a full day’s attendance (Owl Labs, State of Hybrid Work 2025). The broader shift toward flexible work is covered in our remote work statistics. We aggregated data from Kastle Systems, Placer.ai, the Flex Index, WFH Research at Stanford, the US Bureau of Labor Statistics, Gallup, Owl Labs, Robert Half, peer-reviewed journals and the employer announcements listed in the methodology.
TL;DR
- Weekly office occupancy was 54.7% of the early-2020 baseline in late September 2026; the peak day hit 65.3% (Kastle Systems, Back to Work Barometer 2026).
- The post-pandemic weekly record is 56.3%, set in December 2025, with a single-day record of 66.0% (Kastle Systems, December 2025).
- Office visits in August 2026 were 32.5% below August 2019, the smallest August gap since the pandemic (Placer.ai Office Index, September 2026).
- 34% of US firms require full-time office presence, up 2 points in a year; 66% still offer flexibility (Flex Index 2025).
- Required office time rose 12% since early 2024, but actual attendance rose only 1-3% (Flex Index 2025).
- 43% of US hybrid employees sometimes go in for a few hours just to show their faces (Owl Labs, State of Hybrid Work 2025).
- 36% of employers increased required on-site days over the past year (Robert Half 2026).
- 87% of US job postings analyzed in Q2 2026 were fully on-site, up from 65% in Q4 2025 (Robert Half, Q2 2026).
- 21.9% of people at work teleworked in September 2026, 10.7% for all their hours (BLS, Current Population Survey Table A-41).
- About 26% of US paid workdays in September 2026 were work-from-home days (WFH Research, SWAA October 2026).
- 52% of remote-capable US employees work hybrid, 26% fully remote, 22% on-site (Gallup, May 2026).
- Hybrid work cut quit rates by one-third with no effect on performance grades in a 1,612-person trial (Bloom, Han and Liang, Nature 2024).
1. Badge Swipe Data: Offices Are Half Full on Average, Two-Thirds Full on Tuesdays
Badge swipes are the closest thing to a ground-truth measure of mandate compliance, because they count bodies, not policies. In the week that included Tuesday, September 22, 2026, Kastle’s 10-city weekly average stood at 54.7% of its early-2020 baseline, while the Tuesday peak reached 65.3% (Kastle Systems, Back to Work Barometer). That 10.6-point spread between the peak day and the weekly average (65.3 minus 54.7) is the signature of a three-day office week.
Premium buildings tell a different story. Class A+ properties averaged 76.1% and hit 94.0% on their Tuesday peak, so the best space is close to pre-pandemic use midweek while commodity offices lag. The national record was set in December 2025, when the weekly average reached 56.3% and the peak day 66.0% (Kastle Systems, December 2025). Late September 2026 sat 1.6 points below that weekly record, which suggests mandates have stopped producing large new gains.
| Metric | Value | Source |
|---|---|---|
| 10-city weekly average, week of September 22, 2026 | 54.7% (down 1.6 points week over week) | Kastle Systems, Back to Work Barometer 2026 |
| Peak day, Tuesday September 22, 2026 | 65.3% | Kastle Systems, Back to Work Barometer 2026 |
| Class A+ weekly average / Tuesday peak | 76.1% / 94.0% | Kastle Systems, Back to Work Barometer 2026 |
| Highest and lowest cities that week | Dallas 66.3%, Austin 64.7%, New York 64.5%; Philadelphia 40.7% | Kastle Systems, Back to Work Barometer 2026 |
| West Coast cities that week | San Jose 49.6%, San Francisco 47.6%, Los Angeles 47.5% | Kastle Systems, Back to Work Barometer 2026 |
| Post-pandemic weekly record (December 2025) | 56.3% (Class A+ 78.8%) | Kastle Systems, December 2025 |
| Post-pandemic single-day record (December 2025) | 66.0% (Class A+ 95.5%) | Kastle Systems, December 2025 |
| February 2024 benchmark | Weekly 53.0%, peak day 63.1% | Kastle Systems, March 2024 |
Context note: Kastle tracks anonymized keycard, fob and app entries from more than 300,000 users in 10 metro areas against a pre-pandemic baseline set in early 2020. Eight of the 10 cities declined in the September 22 week, a reminder that weekly readings move with holidays, weather and school calendars. The February 2024 figures come from Kastle’s peak day analysis.
2. Foot Traffic and the Friday Problem
Independent location data confirm the plateau-with-a-slow-climb pattern. Office visits in August 2026 were 32.5% below August 2019, the smallest August gap since the pandemic, and up 6.2% from August 2025 (Placer.ai, August 2026 Office Index). Miami (-11.6% vs 2019) and New York (-18.0%) are closest to normal; New York also posted the strongest year-over-year growth at 8.9%.
The week has been reshaped, not just shortened. Kastle’s 2023 analysis found Monday occupancy 21% below the Tuesday-to-Thursday average and Friday 43% below, against gaps of only 1-2% for Monday and 12% for Friday before March 2020 (Kastle Systems, Evidence of the New Hybrid Work Pattern, 2023). That is the practical meaning of most hybrid policies: employees comply on anchor days and treat the edges of the week as remote by default.
| Metric | Value | Source |
|---|---|---|
| National office visits vs 2019, August 2026 | -32.5% (smallest August gap since the pandemic) | Placer.ai Office Index, September 2026 |
| Year-over-year change, August 2026 | +6.2% | Placer.ai Office Index, September 2026 |
| Closest markets to 2019 | Miami -11.6%, New York -18.0%, Dallas -20.8%, Atlanta -29.4% | Placer.ai Office Index, September 2026 |
| Fastest year-over-year growth | New York +8.9%, Boston +8.7% | Placer.ai Office Index, September 2026 |
| Markets that slipped year over year | Washington DC -0.8%, Denver -3.4% | Placer.ai Office Index, September 2026 |
| Monday vs midweek occupancy (2023) | 21% below Tuesday-Thursday (1-2% pre-2020) | Kastle Systems, 2023 |
| Friday vs midweek occupancy (2023) | 43% below Tuesday-Thursday (12% pre-2020) | Kastle Systems, 2023 |
Context note: August 2026 had 21 working days versus 22 in August 2019, which Placer.ai flags when comparing the two. Kastle and Placer.ai use different baselines (early 2020 keycard entries vs 2019 device visits), so their percentages should not be compared directly. The Monday and Friday gaps are the most recent available published breakdown (Kastle Systems, 2023).
3. Mandates on Paper: What Employers Require
Policy has moved further than people. 34% of US firms now require full-time office presence, a 2-point increase from the previous year, while 66% still offer some location flexibility (Flex Index, State of Flexible Work statistics). The largest companies are stricter: 29% of the Fortune 100 require full-time office attendance and 45% require 4 or 5 days, although 71% remain flexible in some form, with a 3-day hybrid schedule the most common policy at 35%.
The high-profile mandates set the tone. Amazon told corporate employees on September 16, 2024 that it would replace its three-day requirement with five days in the office effective January 2, 2025, and asked each senior organization to raise its ratio of individual contributors to managers by at least 15% by the end of Q1 2025 (Amazon, Message from CEO Andy Jassy, 2024). In the public sector, a presidential memorandum dated January 20, 2025 directed agencies to “terminate remote work arrangements” and return employees to their duty stations on a full-time basis, leaving exemptions to agency heads (The White House, Return to In-Person Work, 2025).
| Metric | Value | Source |
|---|---|---|
| US firms requiring full-time office presence | 34% (up 2 points in a year) | Flex Index 2025 |
| US firms offering location flexibility | 66% | Flex Index 2025 |
| Fortune 100 requiring full-time office | 29% | Flex Index 2025 |
| Fortune 100 requiring 4 or 5 days | 45% | Flex Index 2025 |
| Fortune 100 with 3-day hybrid policy | 35% (most common policy) | Flex Index 2025 |
| Employers that increased required on-site days in the past year | 36% | Robert Half, Remote Work Statistics 2026 |
| Amazon corporate policy from January 2, 2025 | 5 days in office (previously 3) | Amazon, September 16, 2024 |
| Federal executive branch, January 20, 2025 memo | Full-time in-person work, exemptions at agency discretion | The White House, 2025 |
Context note: the federal mandate landed on an office estate that was already underused. GAO found 17 of the 24 agencies it reviewed used an estimated average of 25% or less of headquarters capacity in early 2023, while agencies spend about 7 billion dollars a year on office buildings (GAO-23-106200, July 2023). Most recent available data: GAO, 2023.
4. The Compliance Gap: Required Days vs Days Actually Worked On-Site
This is where mandates meet human behavior. Required office time rose 12% since early 2024, but actual attendance increased only 1-3% (Flex Index 2025). Put differently, policy tightened 4 to 12 times faster than behavior changed (12 / 3 and 12 / 1), a ratio that helps explain why Kastle’s late-September 2026 reading still sat below its December 2025 record despite a steady flow of new mandates.
Survey data show how employees close the gap themselves. 43% of US hybrid employees admit they sometimes go to the office for a few hours just to show their faces, and another 12% plan to try it (Owl Labs, State of Hybrid Work 2025). WFH Research finds that employer plans for work from home hover around 1.3 to 1.5 days per week and have resembled actual work-from-home since mid-2022, which suggests hybrid employers largely get the schedule they plan, while the strictest mandates are where slippage concentrates (WFH Research, SWAA October 2026 update).
| Metric | Value | Source |
|---|---|---|
| Change in required office time since early 2024 | +12% | Flex Index 2025 |
| Change in actual attendance over the same period | +1-3% | Flex Index 2025 |
| Hybrid employees who ‘coffee badge’ sometimes | 43% (plus 12% who plan to) | Owl Labs, State of Hybrid Work 2025 |
| Hybrid workers going in 3 / 4 days a week | 39% / 34% | Owl Labs, State of Hybrid Work 2025 |
| Workers who would start job hunting if flexibility were removed | 40% (5% would quit outright, 22% would expect a raise) | Owl Labs, State of Hybrid Work 2025 |
| Employers that had not changed remote or hybrid policy in the past year | 73% | Owl Labs, State of Hybrid Work 2025 |
| Employer-planned WFH days | 1.3 to 1.5 days per week, close to actual WFH since mid-2022 | WFH Research, SWAA October 2026 |
Context note: Owl Labs surveyed 2,000 full-time US workers aged 18 and older in July 2025. Its findings are self-reported; badge data such as Kastle’s cannot distinguish a two-hour visit from a full day, so coffee badging inflates swipe-based compliance figures rather than lowering them. Disengagement under strict policies overlaps with trends in our quiet quitting statistics.
5. How Many People Actually Work From Home in 2026
Every major series agrees that remote work is a minority arrangement that refuses to disappear. 21.9% of people at work in September 2026 teleworked or worked at home for pay: 10.7% for all their hours and 11.2% for some (BLS, Current Population Survey Table A-41). WFH Research estimates that about 26% of paid workdays in September 2026 were work-from-home days, and that over the past year 62% of full-time employees were fully on-site, 26% hybrid and 12% fully remote.
The employer side is moving faster than the worker side. Robert Half’s analysis of job postings found 87% of roles posted in Q2 2026 were fully on-site, up from 65% in Q4 2025, with 10% hybrid and 3% fully remote (Robert Half, Remote Work Statistics and Trends). New hires face stricter terms than incumbents, which means compliance pressure is being built in at the hiring stage instead of renegotiated with existing staff.
| Metric | Value | Source |
|---|---|---|
| People at work who teleworked, September 2026 | 21.9% (34.7 million of 158.5 million) | BLS, CPS Table A-41 |
| Teleworked all hours / some hours | 10.7% / 11.2% | BLS, CPS Table A-41 |
| Employed people working at home on days worked, 2025 | 35% (70% worked at their workplace) | BLS, American Time Use Survey 2025 |
| Share of paid workdays from home, September 2026 | About 26% | WFH Research, SWAA October 2026 |
| Full-time employees on-site / hybrid / fully remote, past year | 62% / 26% / 12% | WFH Research, SWAA October 2026 |
| Remote-capable employees hybrid / remote / on-site, May 2026 | 52% / 26% / 22% | Gallup, Hybrid Work Indicator 2026 |
| Job postings on-site / hybrid / remote, Q2 2026 | 87% / 10% / 3% | Robert Half 2026 |
| US workers in office / hybrid / remote, July 2025 | 63% / 28% / 9% | Owl Labs, State of Hybrid Work 2025 |
Context note: the series measure different things. BLS counts telework during a single reference week, ATUS counts any work at home on a day worked (BLS, American Time Use Survey 2025, released June 25, 2026), and Gallup’s figures cover only remote-capable employees, which is why its hybrid share is roughly double the others (Gallup, Hybrid Work indicator).
6. Who Can Avoid the Office: Industry, Education, Firm Age and Country
Mandates bite hardest where remote work was most common. In September 2026, 53.5% of workers in the information industry and 52.8% in financial activities teleworked, against 7.9% in leisure and hospitality (BLS, CPS Table A-42). In management, business and financial operations occupations, 13,288 thousand of 30,967 thousand people at work teleworked, a rate of 42.9% (13,288 / 30,967). These are the white-collar populations at which most corporate mandates are aimed.
Education is the sharpest divider. Among workers 25 and older, 37.8% with a bachelor’s degree or higher teleworked in September 2026, versus 8.3% of high school graduates with no college (BLS, CPS Table A-41). Firm age matters too: employees at firms founded after 2015 work from home nearly twice as often as those at firms founded before 1990 (Aksoy, Bloom, Cranney, Davis, Dolls and Zarate, NBER 2026). Internationally, college-educated employees in English-speaking countries average 1.3 to 1.9 WFH days per week, while many East Asian countries report less than one (Zarate et al., PNAS 2025).
| Metric | Value | Source |
|---|---|---|
| Telework rate by industry, September 2026 | Information 53.5%, financial activities 52.8%, professional and business services 43.3% | BLS, CPS Table A-42 |
| Lower-telework industries, September 2026 | Government 23.8%, manufacturing 19.2%, leisure and hospitality 7.9% | BLS, CPS Table A-42 |
| Telework by education (25+), September 2026 | Bachelor’s or higher 37.8%; high school, no college 8.3% | BLS, CPS Table A-41 |
| Work at home on days worked by education, 2025 | Bachelor’s or higher 51%; high school, no college 19% | BLS, American Time Use Survey 2025 |
| Telework by sex, September 2026 | Women 23.9%, men 20.1% | BLS, CPS Table A-41 |
| Firms under 500 employees that are fully flexible | 67% (covering half the US workforce) | Flex Index 2025 |
| WFH at firms founded after 2015 vs before 1990 | Nearly twice as frequent | NBER Working Paper 34795 (2026) |
| WFH days per week, English-speaking countries | 1.3 to 1.9 (East Asia often below 1) | G-SWA, PNAS 2025 (14,427 workers, 37 countries) |
Context note: the G-SWA was fielded from November 2024 to February 2025 and covers only full-time, college-educated workers. The NBER paper uses about 76,716 SWAA responses from January 2024 to December 2025.
7. Outcomes: What Attendance Mandates Buy and What They Cost
The research case for strict mandates remains thin. A six-month randomized trial of 1,612 employees found hybrid work (two days a week at home) reduced quit rates by one-third, with no effect on performance grades or promotions over the next two years (Bloom, Han and Liang, Nature 2024). The 395 managers in that trial moved from expecting a -2.6% productivity effect to perceiving +1.0% after the experiment. Most recent available randomized evidence: Nature, 2024.
Newer observational work points the same way. A 2026 study of 7,704 employees found the highest well-being among remote workers (mean 4.22) versus hybrid (4.12) and onsite (3.89), with one-year turnover of 7.8% for remote and 8.8% for onsite staff, a turnover difference that was not statistically significant (Lezcano et al., Frontiers in Psychology 2026). A preprint posted October 3, 2026 used 13,956 employee-month badge-swipe records for 1,163 employees at a multinational financial firm and found that a faster increase in attendance predicted lower performance, while higher attendance predicted more stress and burnout (Luan, Kim, Lyu and Toh, PsyArXiv 2026). Those burnout signals connect to the trends in our workplace burnout statistics.
| Metric | Value | Source |
|---|---|---|
| Quit-rate effect of hybrid work (RCT, 1,612 employees) | Reduced by one-third | Bloom, Han and Liang, Nature 2024 |
| Performance and promotions over two years | No difference | Bloom, Han and Liang, Nature 2024 |
| Manager view of hybrid productivity, before vs after | -2.6% to +1.0% | Bloom, Han and Liang, Nature 2024 |
| Well-being score: remote / hybrid / onsite | 4.22 / 4.12 / 3.89 (n = 7,704) | Lezcano et al., Frontiers in Psychology 2026 |
| One-year turnover: remote vs onsite | 7.8% vs 8.8% (not significant) | Lezcano et al., Frontiers in Psychology 2026 |
| Badge-swipe study sample | 13,956 employee-months, 1,163 employees | Luan et al., PsyArXiv preprint 2026 |
| Revenue growth, fully flexible vs mandate-driven firms, 2019-2024 | 1.7x faster | Flex Index 2025 |
| Engaged employees: hybrid vs on-site remote-capable, May 2026 | 34% vs 29% | Gallup, Hybrid Work Indicator 2026 |
Context note: the Flex Index revenue comparison is correlational and does not establish that flexibility causes growth; firms that choose flexibility may differ in sector and maturity. The Luan et al. paper is a preprint that has not completed peer review. The Bloom trial was run at a Chinese technology company in 2021-2022. Earlier context on attendance trends is collected in our return to office statistics.
Summary: Hybrid Work Mandate Compliance by the Numbers
| Metric | Value | Source |
|---|---|---|
| Weekly office occupancy, late September 2026 | 54.7% of early-2020 baseline | Kastle Systems, Back to Work Barometer 2026 |
| Peak day occupancy, Tuesday September 22, 2026 | 65.3% | Kastle Systems, Back to Work Barometer 2026 |
| Class A+ peak day | 94.0% | Kastle Systems, Back to Work Barometer 2026 |
| Post-pandemic weekly record | 56.3% (December 2025) | Kastle Systems, December 2025 |
| Office visits vs 2019, August 2026 | -32.5% | Placer.ai Office Index, September 2026 |
| Friday vs midweek occupancy | 43% below (2023) | Kastle Systems, 2023 |
| US firms requiring full-time office | 34% | Flex Index 2025 |
| Fortune 100 requiring 4 or 5 days | 45% | Flex Index 2025 |
| Required office time vs actual attendance since early 2024 | +12% vs +1-3% | Flex Index 2025 |
| Hybrid employees who coffee badge | 43% | Owl Labs, State of Hybrid Work 2025 |
| Employers that raised required on-site days | 36% | Robert Half 2026 |
| Fully on-site job postings | 87% in Q2 2026 (65% in Q4 2025) | Robert Half 2026 |
| People at work who teleworked | 21.9% (September 2026) | BLS, CPS Table A-41 |
| Paid workdays from home | About 26% (September 2026) | WFH Research, SWAA October 2026 |
| Full-time employees on-site / hybrid / remote | 62% / 26% / 12% | WFH Research, SWAA October 2026 |
| Remote-capable employees in hybrid arrangements | 52% | Gallup, May 2026 |
| Information industry telework rate | 53.5% | BLS, CPS Table A-42 |
| Quit-rate effect of hybrid work | Down one-third | Bloom, Han and Liang, Nature 2024 |
| Federal agencies at 25% or less of HQ capacity | 17 of 24 (early 2023) | GAO-23-106200 |
Methodology and Sources
Every figure above was read during research for this article in a primary dataset, a government release, a company announcement, a peer-reviewed paper or an organization’s published survey. Badge, location and survey data measure different things, so we report each series on its own terms and do not mix baselines.
- Kastle Systems: Back to Work Barometer (weekly), all-time post-pandemic highs, December 2025, peak day data, March 2024, evidence of the new hybrid work pattern, September 2023
- Placer.ai: August 2026 Office Index (September 14, 2026)
- Flex Index (Scoop): State of Flexible Work statistics
- WFH Research (Barrero, Bloom, Davis): SWAA monthly update, October 2026, NBER paper on firm and CEO age (2026), G-SWA cross-country paper, PNAS 2025
- US Bureau of Labor Statistics: CPS Table A-41, telework by demographics, September 2026, CPS Table A-42, telework by occupation and industry, September 2026, American Time Use Survey 2025 (June 25, 2026)
- Gallup: Hybrid Work indicator (May 2026 data)
- Owl Labs: State of Hybrid Work 2025, US report
- Robert Half: Remote Work Statistics and Trends (Q2 2026 job postings, TalentNeuron data)
- Amazon: Message from CEO Andy Jassy, September 16, 2024
- The White House: Presidential Memorandum, Return to In-Person Work, January 20, 2025
- US Government Accountability Office: GAO-23-106200, federal building utilization, July 2023
- Peer-reviewed and preprint research via Europe PMC: Bloom, Han and Liang, Nature 2024, Lezcano et al., Frontiers in Psychology 2026, Luan et al., PsyArXiv 2026
- Data watch: Sources disagree on the size of the remote and hybrid population because they measure different groups: BLS counts all people at work in one reference week (21.9% teleworked), WFH Research counts full-time wage and salary employees (26% hybrid, 12% fully remote), Owl Labs surveys 2,000 full-time workers (28% hybrid, 9% remote) and Gallup covers only remote-capable employees (52% hybrid). Robert Half’s on-site posting share jumped from 65% (Q4 2025) to 87% (Q2 2026); its page attributes the postings to TalentNeuron data mapped with a large language model, so part of that change may reflect methodology. The Flex Index statistics page reflects its 2025 reporting and its full reports page could not be loaded (server error and rate limits), so the date of each Flex Index figure could not be confirmed beyond 2025. The Kastle weekly page is updated in place, and the figures here are for the week including Tuesday, September 22, 2026. Kastle’s day-of-week gaps (2023), GAO’s federal building utilization (2023) and the Bloom trial (2021-2022 fieldwork, published 2024) are the most recent available data of their kind. The Luan et al. study is a preprint. Census American Community Survey commuting tables could not be queried without an API key and were not used. A JPMorgan Chase five-day policy announcement could not be verified on a primary page and was excluded.
Last updated: October 3, 2026. We update this roundup quarterly, and the next refresh is expected when WFH Research publishes its November 2026 SWAA update and BLS releases October 2026 telework data in the Current Population Survey.