Return to Office Statistics 2026
By Axis Intelligence Research
Co-author: David Park | Last updated: August 17, 2026 | License: CC BY 4.0
The U.S. telework rate was 22.2% in July 2026 — 34.1 million people. That is 1.6 percentage points below the October 2024 peak of 23.8%, and 4.7 points above where the series began in October 2022. The return to office is real, measurable, and far smaller than four years of headlines implied.
Quick Answer
Roughly one in five American workers still teleworks in any given week. In July 2026, 22.2% of people at work teleworked or worked at home for pay: 11.2% for some of their hours, 11.0% for all of them. The Axis Office Reabsorption Index (ORI™) — Axis Intelligence Research’s measure of how far the office has pulled work back — reads 66.9 for July 2026 on a 0–100 scale, against annual readings of 67.2 for 2024 and 67.8 for 2025. Reabsorption has stalled, not accelerated.
Key Findings
- Axis Intelligence Research finds the U.S. telework rate peaked at 23.8% in October 2024, fell to 21.5% by May 2025 after the federal return-to-office directive, and had recovered to 22.6% by March 2026 — retracing 47.8% of the decline.
- Axis Intelligence Research finds federal government workers post an Office Reabsorption Index of 71.4 in July 2026, but local government workers score higher at 78.8 — the most office-bound employer class in the United States, and it got there without a mandate.
- Axis Intelligence Research finds a 42.7-point Office Reabsorption Index spread between accommodation and food services (84.8) and professional and technical services (42.1) in July 2026 — the widest gap between any two U.S. industries.
- Axis Intelligence Research finds Arkansas is the most office-reabsorbed U.S. jurisdiction (ORI 73.7) and the District of Columbia the least (51.3), a 22.4-point spread across 50 states and D.C.
- Axis Intelligence Research finds the hybrid conversion rate — the share of teleworkers who also spend part of the week on site — fell from 53.1% in 2025 to 50.5% in July 2026, meaning the workers who remain remote are becoming more remote, not less.
What Percentage of Workers Are Back in the Office in 2026?
In July 2026, 119.3 million of the 153.4 million people at work did not telework or work at home for pay at all. That is 77.8% fully on site — the plain answer to the question most people are actually asking, and the number that has barely moved in four years.
The Bureau of Labor Statistics telework series began in October 2022 at a 17.9% telework rate. It climbed through 2023 and 2024, peaked at 23.8% in October 2024, and has since settled into a narrow band. BLS itself notes the rate has ranged between 21.5% and 23.0% over the year to March 2026.
The telework rate month by month, 2022–2026
| Period | Telework rate | Some hours | All hours | On site | Source |
|---|---|---|---|---|---|
| Oct 2022 (series start) | 17.9% | — | — | — | BLS CPS |
| Oct 2024 (series peak) | 23.8% | — | — | — | BLS CPS |
| Feb 2025 | 23.7% | — | — | — | BLS CPS |
| May 2025 (post-directive low) | 21.5% | — | — | — | BLS CPS |
| Mar 2026 | 22.6% | — | — | — | BLS CPS |
| Jul 2026 | 22.2% | 11.2% | 11.0% | 77.8% | BLS CPS Table A-42 |
| 2024 annual average | 22.8% | 11.9% | 10.9% | 77.2% | BLS LAU |
| 2025 annual average | 22.4% | 11.9% | 10.5% | 77.6% | BLS CPS Table 59 |
October 2025 data are unavailable because of the 2025 lapse in appropriations. Monthly figures are not seasonally adjusted.
Fully remote versus hybrid versus on site
Two independent surveys, built on different methodologies, land within a percentage point of each other on the fully remote share — which is the strongest evidence available that the number is real.
BLS puts workers who teleworked all their hours at 11.0% of people at work in July 2026. The academic Survey of Working Arrangements and Attitudes, run by Barrero, Bloom, Buckman and Davis across 38,737 full-time wage and salary employees between June 2025 and May 2026, puts fully remote at 12%, hybrid at 26%, and fully on site at 62%.
Axis Intelligence Research does not average these two figures. The BLS denominator is everyone at work in a single reference week; the SWAA denominator is full-time wage and salary employees over twelve months. Merging them would produce a number that means nothing. Published side by side, they mean a great deal: two research teams, two sampling frames, one answer of roughly one in eight.
David Park, Careers & Workforce: The fear I hear most often, usually late and usually from someone with a mortgage priced around a remote comp band, is that the flexible job market has closed. It has not. It has narrowed and it has stopped growing. Those are different problems, and only one of them requires you to move. Requisitions offering some remote arrangement still cover better than a fifth of everyone at work in any given week, and the on-site share has moved 0.6 percentage points across two full annual averages. If you are planning around attrition risk or a relocation, plan around a flat market, not a closing one.
What Is the Axis Office Reabsorption Index (ORI™)?
Telework rates answer how many people work from home. They do not answer the question employers, journalists and workers actually argue about: how far has the office pulled work back?
Those are different questions, because a workforce can be office-bound two ways. Most of it may never have left — the structural component. Or the people who did leave may now be splitting the week — the behavioural component. A single telework percentage collapses both into one figure and hides which one is moving.
The Axis Office Reabsorption Index (ORI™) separates them.
The formula
ORI = (0.60 × on-site share) + (0.40 × hybrid conversion rate)
Where:
- On-site share = the percentage of people at work who did not telework or work at home for pay
- Hybrid conversion rate = (teleworked some hours ÷ total teleworked) × 100
Both inputs come from the same BLS table and the same reference period, so the methodologies are compatible by construction. The index runs 0–100. Higher means more office-reabsorbed.
The 60/40 weighting places structural on-site presence ahead of behavioural conversion, because presence is the direct measure and conversion is the marginal signal. The weights are fixed and disclosed so any competent outsider can recompute every reading in this article from the linked BLS tables.
Worked example: federal government, July 2026
- On-site share = 83.6%
- Hybrid conversion = 8.7 ÷ 16.4 × 100 = 53.0488%
- 0.60 × 83.6 = 50.16
- 0.40 × 53.0488 = 21.2195
- ORI = 71.4
National readings
| Reading | ORI | On-site share | Hybrid conversion | Basis |
|---|---|---|---|---|
| 2024 | 67.2 | 77.2% | 52.2% | Annual average |
| 2025 | 67.8 | 77.6% | 53.1% | Annual average |
| July 2026 | 66.9 | 77.8% | 50.5% | Single month |
The 2024 and 2025 readings are directly comparable: both are annual averages, and reabsorption rose 0.6 points between them. The July 2026 reading is a single unadjusted month and is reported separately rather than spliced onto the annual series.
What that monthly reading shows is worth sitting with. The on-site share kept edging up — 77.2%, 77.6%, 77.8%. But hybrid conversion fell from 53.1% to 50.5%. Fewer of the remaining teleworkers are splitting their week. The office is holding the people who never left and losing ground with the people who did.
Scope of the index
The ORI measures position, not policy. It cannot tell you whether a segment arrived at its reading through a mandate, a lease decision, or the simple fact that the work requires hands on equipment. In segments where telework is rare, the hybrid conversion component rests on a small base and moves erratically — accommodation and food services reports a 67.6% conversion rate drawn from a 3.7% telework rate. The index is most informative in the remote-capable band, where both components carry weight. State readings use 2024 annual averages, the most recent state-level release; national and industry readings use July 2026.
Which Industries Have Returned to the Office the Most?
The industry table is where the return-to-office story stops being one story.
| Industry | Telework rate | Some hours | All hours | On site | ORI | Source |
|---|---|---|---|---|---|---|
| Accommodation and food services | 3.7% | 2.5% | 1.2% | 96.3% | 84.8 | BLS A-42 |
| Mining, quarrying, oil and gas | 11.4% | 7.9% | 3.5% | 88.6% | 80.9 | BLS A-42 |
| Construction | 10.1% | 6.4% | 3.7% | 89.9% | 79.3 | BLS A-42 |
| Leisure and hospitality | 7.5% | 4.3% | 3.2% | 92.5% | 78.4 | BLS A-42 |
| Transportation and warehousing | 8.3% | 4.7% | 3.6% | 91.7% | 77.7 | BLS A-42 |
| Retail trade | 10.6% | 5.6% | 5.0% | 89.4% | 74.8 | BLS A-42 |
| Manufacturing | 19.1% | 11.6% | 7.5% | 80.9% | 72.8 | BLS A-42 |
| Utilities | 23.4% | 15.3% | 8.1% | 76.6% | 72.1 | BLS A-42 |
| Public administration | 22.1% | 13.9% | 8.2% | 77.9% | 71.9 | BLS A-42 |
| Other services | 18.1% | 10.0% | 8.1% | 81.9% | 71.2 | BLS A-42 |
| Arts, entertainment, recreation | 18.8% | 9.6% | 9.1% | 81.2% | 69.1 | BLS A-42 |
| Health care and social assistance | 18.4% | 8.9% | 9.4% | 81.6% | 68.3 | BLS A-42 |
| Wholesale trade | 23.8% | 13.0% | 10.8% | 76.2% | 67.6 | BLS A-42 |
| Educational services | 22.3% | 11.6% | 10.7% | 77.7% | 67.4 | BLS A-42 |
| Management, admin and waste services | 16.5% | 6.9% | 9.6% | 83.5% | 66.8 | BLS A-42 |
| Real estate and rental and leasing | 40.6% | 24.1% | 16.5% | 59.4% | 59.4 | BLS A-42 |
| Information | 49.0% | 21.3% | 27.7% | 51.0% | 48.0 | BLS A-42 |
| Finance and insurance | 55.4% | 26.1% | 29.3% | 44.6% | 45.6 | BLS A-42 |
| Professional and technical services | 57.9% | 24.4% | 33.4% | 42.1% | 42.1 | BLS A-42 |
July 2026, not seasonally adjusted. Components may not sum exactly because of BLS rounding.
The three industries that never came back
Professional and technical services, finance and insurance, and information sit at ORI readings of 42.1, 45.6 and 48.0. In all three, more workers are remote in some form than are fully on site. Professional and technical services is the outlier of outliers: 33.4% of its workforce teleworked every hour they worked in July 2026, a rate nine times the entire telework rate of accommodation and food services.
BLS’s June 2026 industry snapshot put finance and insurance at 59.0% and professional and technical services at 55.8% — one month earlier and within a few points of the July readings, which is what a stable series looks like.
One caution the aggregators skip: BLS notes that effective with January 2025 data, industries reflect the 2022 Census industry classification system, and 2025 figures are not strictly comparable with earlier years. Any industry-level “since 2023” comparison you see quoted this year is crossing a classification break. Axis Intelligence Research does not publish one.
David Park: Run the utilisation maths on the finance number for a moment. Finance and insurance has 7.4 million people at work and 4.1 million of them teleworked in July. If an employer in that band writes a five-day policy, it is not adjusting a schedule — it is repricing the arrangement for more than half its headcount at once, against comp bands that were set when the arrangement was assumed. That is why the announced policies and the measured rates keep diverging. The announcement is cheap. The attrition it prices in is not.
Did the Federal Return-to-Office Mandate Work?
On January 20, 2025, a presidential memorandum published in the Federal Register directed executive branch agency heads to terminate remote work arrangements and require employees to return to in-person work full time, with exemptions at agency discretion. It is the single largest deliberate return-to-office action in U.S. history by headcount.
The measured result, eighteen months later:
| Class of worker | Telework rate | On site | People at work | ORI | Source |
|---|---|---|---|---|---|
| Local government | 12.6% | 87.4% | 8,045k | 78.8 | BLS A-42 |
| Federal government | 16.4% | 83.6% | 3,672k | 71.4 | BLS A-42 |
| State government | 27.5% | 72.5% | 6,054k | 68.7 | BLS A-42 |
| Private industries | 20.9% | 79.1% | 120,211k | 67.6 | BLS A-42 |
| Self-employed, unincorporated | 32.6% | 67.4% | 9,161k | 55.3 | BLS A-42 |
| Self-employed, incorporated | 42.7% | 57.3% | 6,183k | 55.0 | BLS A-42 |
Federal workers are more office-bound than private-sector workers by 3.8 index points. The mandate moved the number.
It did not move it to the top. Local government scores 78.8 — 7.4 points above federal — and no memorandum produced that. County clerks, transit operators, school district staff and municipal maintenance crews were on site all along. The most office-bound employer class in America is the one nobody wrote an executive action about.
What the Defense Department’s own numbers show
The clearest agency-level accounting comes from the Government Accountability Office, which reported in January 2026 that as of July 2025 about 8% of Department of Defense employees — 62,000 of 780,000 — had not returned to in-person work. Of those, 45,000 were in deferred resignation status or held other exemptions, and 17,000 had reasonable accommodations.
GAO also found DOD’s underlying eligibility data incomplete: between 65% and 68% of DOD civilian positions were telework-eligible on paper, but the department could not reliably say how many employees actually were. In May 2024 DOD publicly reported 61,549 remote employees; one month later it told GAO the figure was 35,558.
That discrepancy deserves more attention than it received. A department that cannot count its own remote workers to within 26,000 is a department whose compliance reporting should be read as a direction, not a measurement.
The recovery nobody reported
Here is the finding that the twelve months of return-to-office coverage missed, and it comes straight out of the BLS monthly series.
The national telework rate peaked at 23.8% in October 2024, three months before the directive. It read 23.7% in February 2025, then fell for three consecutive months to 21.5% in May 2025 — a 2.3-point decline. By March 2026 it was back to 22.6%.
Axis Intelligence Research calculates that the market retraced 47.8% of the post-directive decline within ten months. The federal action removed roughly one worker in forty-three from the telework column nationally, and the private sector put half of them back.
Which States Are Most and Least Back in the Office?
State-level telework data is published annually by BLS through the Local Area Unemployment Statistics program and is, as far as Axis Intelligence Research can establish, almost never analysed. The 2024 annual averages are the most recent release.
Ranked by the Axis Office Reabsorption Index, all 50 states and the District of Columbia:
| Rank | Jurisdiction | ORI | Telework rate | On site | Rank | Jurisdiction | ORI | Telework rate | On site |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Arkansas | 73.7 | 11.6% | 88.4% | 27 | Delaware | 67.3 | 22.4% | 77.6% |
| 2 | North Dakota | 73.5 | 14.3% | 85.7% | 28 | New Jersey | 67.2 | 27.1% | 72.9% |
| 3 | Iowa | 73.3 | 15.7% | 84.3% | 29 | Minnesota | 66.9 | 25.9% | 74.1% |
| 4 | Hawaii | 73.1 | 15.8% | 84.2% | 30 | California | 66.8 | 25.5% | 74.5% |
| 5 | Louisiana | 72.6 | 12.8% | 87.2% | 31 | Tennessee | 66.8 | 19.5% | 80.5% |
| 6 | Alaska | 72.5 | 16.8% | 83.2% | 32 | Idaho | 66.7 | 19.7% | 80.3% |
| 7 | South Dakota | 72.3 | 14.6% | 85.4% | 33 | Nevada | 66.5 | 16.1% | 83.9% |
| 8 | Oklahoma | 71.9 | 14.7% | 85.3% | 34 | Pennsylvania | 66.4 | 23.8% | 76.2% |
| 9 | West Virginia | 71.8 | 12.3% | 87.7% | 35 | Illinois | 66.3 | 26.3% | 73.7% |
| 10 | Indiana | 71.7 | 18.2% | 81.8% | 36 | North Carolina | 66.3 | 22.0% | 78.0% |
| 11 | Mississippi | 71.5 | 7.3% | 92.7% | 37 | Maine | 66.2 | 23.5% | 76.5% |
| 12 | Kentucky | 71.4 | 15.6% | 84.4% | 38 | Texas | 65.7 | 22.5% | 77.5% |
| 13 | New York | 71.4 | 21.2% | 78.8% | 39 | Georgia | 65.6 | 22.3% | 77.7% |
| 14 | Wyoming | 70.9 | 13.5% | 86.5% | 40 | Vermont | 65.5 | 26.3% | 73.7% |
| 15 | Connecticut | 70.6 | 25.2% | 74.8% | 41 | Massachusetts | 65.2 | 31.4% | 68.6% |
| 16 | Nebraska | 70.2 | 18.2% | 81.8% | 42 | Florida | 64.8 | 19.0% | 81.0% |
| 17 | Kansas | 70.1 | 20.3% | 79.7% | 43 | New Hampshire | 64.1 | 26.0% | 74.0% |
| 18 | Rhode Island | 70.1 | 23.4% | 76.6% | 44 | Virginia | 64.0 | 28.6% | 71.4% |
| 19 | Alabama | 70.0 | 10.7% | 89.3% | 45 | Utah | 63.6 | 26.8% | 73.2% |
| 20 | Michigan | 69.7 | 21.2% | 78.8% | 46 | Maryland | 63.2 | 30.4% | 69.6% |
| 21 | New Mexico | 69.1 | 17.4% | 82.6% | 47 | Washington | 62.7 | 30.8% | 69.2% |
| 22 | Ohio | 68.6 | 21.3% | 78.7% | 48 | Arizona | 61.8 | 27.5% | 72.5% |
| 23 | Missouri | 68.4 | 21.8% | 78.2% | 49 | Oregon | 61.7 | 29.3% | 70.7% |
| 24 | Wisconsin | 68.4 | 21.9% | 78.1% | 50 | Colorado | 61.1 | 32.4% | 67.6% |
| 25 | Montana | 68.2 | 20.6% | 79.4% | 51 | District of Columbia | 51.3 | 56.8% | 43.2% |
| 26 | South Carolina | 68.2 | 18.2% | 81.8% | National | 67.2 | 22.8% | 77.2% |
Source: BLS Local Area Unemployment Statistics, 2024 annual averages. ORI calculated by Axis Intelligence Research. [Older data] — 2024 is the most recent state-level telework release; BLS publishes these figures annually.
Median jurisdiction: 68.2. National: 67.2. Spread: 22.4 points.
New York ranks 13th — and that is not a mistake
New York State posts an ORI of 71.4 on a 21.2% telework rate. It ranks alongside Kentucky, ahead of Michigan and Ohio, and well ahead of California at 30th. The reason is the hybrid conversion component: 12.8 of New York’s 21.2 telework points are workers doing some hours at home, a 60.4% conversion rate. New Yorkers who telework mostly still commute. New York State also is not New York City, and the state figure blends Buffalo, Syracuse and the Southern Tier into the Manhattan number.
Colorado, Oregon, Washington and Arizona cluster at the bottom of the mainland ranking on the opposite pattern: high telework rates weighted toward fully remote arrangements.
The one jurisdiction the federal mandate was aimed at
The District of Columbia is not close to any other jurisdiction. Its 2024 telework rate was 56.8% — more than double the national 22.8% — and its ORI of 51.3 sits 9.8 points below Colorado, the next lowest.
The date matters more than the number. These are 2024 annual averages: the last full pre-directive reading for the jurisdiction the January 2025 memorandum landed on hardest. Maryland (30.4%) and Virginia (28.6%) ranked 46th and 44th on the same table. Three of the six least office-reabsorbed jurisdictions in America form a contiguous triangle around the federal workforce, measured the year before that workforce was ordered back.
When BLS publishes the 2025 state table, that triangle is where the change will show up. Axis Intelligence Research will publish the revised ORI rankings within one week of that release.
Which Occupations Are Still Working Remotely?
Occupation predicts telework more sharply than industry, state or employer type.
| Occupation | Telework rate | Some hours | All hours | On site | Source |
|---|---|---|---|---|---|
| Computer and mathematical | 62.5% | 26.4% | 36.1% | 37.5% | BLS A-42 |
| Business and financial operations | 53.9% | 25.0% | 28.9% | 46.1% | BLS A-42 |
| Legal | 50.8% | 31.4% | 19.4% | 49.2% | BLS A-42 |
| Arts, design, entertainment, sports, media | 48.6% | 21.0% | 27.6% | 51.4% | BLS A-42 |
| Community and social services | 39.8% | 25.6% | 14.3% | 60.2% | BLS A-42 |
| Management | 37.0% | 20.4% | 16.6% | 63.0% | BLS A-42 |
| Architecture and engineering | 35.6% | 24.5% | 11.1% | 64.4% | BLS A-42 |
| Life, physical and social science | 34.9% | 22.0% | 12.9% | 65.1% | BLS A-42 |
| Office and administrative support | 24.0% | 10.4% | 13.6% | 76.0% | BLS A-42 |
| Sales and related | 23.5% | 12.4% | 11.1% | 76.5% | BLS A-42 |
| Education, training and library | 19.1% | 8.3% | 10.8% | 80.9% | BLS A-42 |
| Healthcare support | 12.8% | 2.9% | 9.9% | 87.2% | BLS A-42 |
| Healthcare practitioners and technical | 12.0% | 6.5% | 5.4% | 88.0% | BLS A-42 |
| Personal care and service | 9.3% | 3.1% | 6.1% | 90.7% | BLS A-42 |
| Protective service | 7.6% | 3.9% | 3.7% | 92.4% | BLS A-42 |
| Production | 5.8% | 3.2% | 2.6% | 94.2% | BLS A-42 |
| Installation, maintenance and repair | 5.0% | 2.4% | 2.7% | 95.0% | BLS A-42 |
| Farming, fishing and forestry | 3.8% | 1.4% | 2.4% | 96.2% | BLS A-42 |
| Building and grounds cleaning | 3.3% | 2.3% | 0.9% | 96.7% | BLS A-42 |
| Transportation and material moving | 2.9% | 1.8% | 1.1% | 97.1% | BLS A-42 |
| Construction and extraction | 2.5% | 2.3% | 0.3% | 97.5% | BLS A-42 |
| Food preparation and serving | 0.9% | 0.7% | 0.3% | 99.1% | BLS A-42 |
Computer and mathematical occupations telework at 62.5%; food preparation and serving at 0.9%. That is a 69-fold gap, and it is the single widest disparity in the entire dataset. It also explains most of the age pattern in the monthly series — workers aged 16 to 24 teleworked at 6.7% in March 2026 against 25.1% for those aged 25 to 54, because entry-level roles cluster in the occupations at the bottom of this table. Our Gen Z workforce data covers what that cohort says it wants from flexible arrangements.
Legal occupations are worth a second look. At a 50.8% telework rate, 31.4 points are hybrid and only 19.4 fully remote — the most strongly hybrid-tilted profession in the table. Law firms did not go remote; they went to three days.
Education is the sharpest demographic divide
The American Time Use Survey for 2025, released June 25, 2026 from roughly 6,100 time diaries, measures this from a different angle: on days they worked, 51% of employed people aged 25 and over with a bachelor’s degree or higher did some work at home, against 19% of those with a high school diploma and no college — a 32-point gap.
ATUS also puts the overall figure at 35% doing some work at home on days worked and 70% doing some work at their workplace, with employed women at 38% and men at 31%. The percentages exceed 100 because hybrid days count in both columns.
SWAA tracks the same divide as a share of paid workdays and shows it narrowing at the top:
| Education group | 2024 | 2025 | 2026 |
|---|---|---|---|
| Graduate degree | 36.0% | 34.5% | 31.4% |
| 4-year college degree | 33.8% | 33.2% | 31.0% |
| 1 to 3 years of college | 25.1% | 23.9% | 22.8% |
| High school or less | 20.3% | 20.8% | 19.0% |
Source: WFH Research SWAA, June 2026 update. 2026 averages the months available at release.
Graduate degree holders gave up 4.6 points of work-from-home share across two years. Workers with a high school education or less gave up 1.3. The return to office has been, disproportionately, a return of the highly credentialed. That is the same population facing the largest reskilling requirement this decade, a pressure quantified in our AI adoption and workforce data.
David Park: If you are advising someone on a career move right now, the occupation row matters more than the employer’s stated policy. A computer and mathematical role carries a 62.5% telework rate across the whole economy — that is not one company’s benefit, it is the ambient condition of the job family, and it survives individual employers changing their minds. An office and administrative support role carries 24.0%. Two people can leave the same building under the same policy and face completely different re-hire terms. Check the occupation band before you check the job posting. And check it against what is happening to hiring in that band — the occupations with the highest telework rates are also the ones absorbing the most pressure from automation, which we cover in AI job displacement statistics.
How Many Days a Week Do Hybrid Workers Come In?
Employer plans have been remarkably stable. SWAA has tracked planned work-from-home days since July 2020 and reports employer plans hovering at 1.3 to 1.5 days per week, closely tracking actual work-from-home behaviour since mid-2022. About 25% of paid full days in the United States in May 2026 were work-from-home days.
BLS measures the same thing in hours. Across 2025, teleworkers averaged 26.5 hours per week working at home. Split by type: those who teleworked some hours averaged 17.3 hours at home, while those who teleworked all hours averaged 37.0.
Seventeen hours is a little over two working days. The three-day office week is not a compromise anyone designed. It is where the numbers settled.
Fully remote arrangements are a separate market with separate rules, and a growing share of it is location-independent rather than merely home-based — we track that population in our digital nomad statistics.
How Does U.S. Return to Office Compare Internationally?
Cross-country comparison is harder than it looks, because national statistical offices do not use a common definition — some measure “usually works from home,” others “worked from home in the last seven days,” others full paid days.
Great Britain’s Office for National Statistics found that 29% of workers aged 30 and over followed a hybrid pattern between April and June 2024, against 19% of those aged 16 to 29. Degree holders were more than ten times as likely to work hybrid as those with no qualifications — 42% against 4%. [Older data] — this remains the most recent ONS breakdown of hybrid working by characteristic.
The structural pattern matches the U.S. precisely: age, education, and occupation determine location flexibility, and the gaps are enormous. Axis Intelligence Research does not convert these figures to a common scale, because the underlying survey questions differ enough that a harmonised number would be a construction rather than a measurement.
Methodology
Collection. Every figure in this article was fetched from a primary source during production on August 15, 2026 (published August 17, 2026), and logged with its URL and retrieval date in the accompanying CSV. No figure was drawn from recollection, from a secondary aggregator, or from another publication’s summary of a primary source.
Primary sources used. BLS Current Population Survey Table A-42 (July 2026, published August 7, 2026); BLS CPS Table 59 and Table 63 (2025 annual averages); BLS Local Area Unemployment Statistics state telework table (2024 annual averages); BLS The Economics Daily releases of April 16, 2026 and July 29, 2026; BLS American Time Use Survey 2025 results (USDL-26-1022); the Federal Register publication of the January 20, 2025 presidential memorandum; GAO-26-107601; the June 2026 SWAA update from WFH Research; and the UK Office for National Statistics.
The Axis Office Reabsorption Index. ORI = (0.60 × on-site share) + (0.40 × hybrid conversion rate). On-site share is the BLS-published percentage of people at work who did not telework or work at home for pay. Hybrid conversion rate is BLS-published teleworked-some-hours divided by BLS-published total teleworked, expressed as a percentage. Both components are taken from a single BLS table for a single reference period; no cross-survey inputs are combined. Weights are fixed at 0.60/0.40 and will not change without a stated re-baselining note. All 78 index readings in this dataset were computed twice in independent implementations — one using floating-point arithmetic, one using exact rational arithmetic — and reconciled to one decimal place before publication.
Comparability. Annual averages are compared only with annual averages. The July 2026 monthly reading is reported separately and is not spliced onto the annual ORI series. BLS notes that effective with January 2025 data, industry categories reflect the 2022 Census industry classification system and 2025 figures are not strictly comparable with earlier years; this article therefore publishes no industry time series crossing that break. October 2025 data are absent from the monthly series because of the 2025 lapse in appropriations. Monthly CPS telework figures are not seasonally adjusted.
Scope. The CPS telework question is asked only of people employed and at work during the reference week; it excludes those absent the entire week. It does not distinguish main from second jobs for multiple jobholders. The ORI describes position rather than cause. Where telework is rare, the hybrid conversion component rests on a small base and should be read with that in mind.
About This Dataset
Contents. 1,106 rows covering the U.S. telework rate from October 2022 to July 2026 by month and demographic group; telework status by industry, occupation and class of worker for July 2026; annual averages for 2024 and 2025; state-level telework status for all 50 states and the District of Columbia; American Time Use Survey location-of-work figures; and every Axis Office Reabsorption Index reading with its disclosed formula.
Structure. One row per observation. Every row carries metric, value, unit, as_of_date, geography, segment, source_org, source_document, source_url, retrieved_date, is_primary, axis_calculated, method_note and license. Rows where axis_calculated = yes carry the full formula in method_note.
Coverage. United States (national, 50 states, District of Columbia) and Great Britain. Temporal coverage 2021 to July 2026.
License. CC BY 4.0. Free to use, republish, chart and redistribute, including commercially, with attribution.
Attribution line: Axis Intelligence Research, Return to Office Statistics 2026, 2026.
Related datasets. This page sits in the Axis Intelligence Research statistics library, where every dataset ships with source columns and a CC BY 4.0 licence.
Cite This Research
APA Axis Intelligence Research. (2026). Return to office statistics 2026: Telework rates by industry, state, and occupation. https://axis-intelligence.com/return-to-office-statistics/
MLA Axis Intelligence Research. “Return to Office Statistics 2026: Telework Rates by Industry, State, and Occupation.” Axis Intelligence, 17 Aug. 2026, axis-intelligence.com/return-to-office-statistics/.
Chicago Axis Intelligence Research. “Return to Office Statistics 2026: Telework Rates by Industry, State, and Occupation.” Axis Intelligence. August 17, 2026. https://axis-intelligence.com/return-to-office-statistics/.
Frequently Asked Questions
Does the BLS telework rate count hybrid workers, or only the fully remote?
Both, and this is the most common misreading of the series. The CPS question asks whether a person teleworked or worked at home for pay at any time during the reference week. Someone who worked from home on a single Friday is counted. The 22.2% July 2026 rate splits into 11.2% who teleworked some hours and 11.0% who teleworked all hours — so roughly half the “remote workforce” in the headline number is commuting most of the week.
Why do Gallup and BLS report such different remote work numbers?
Denominators, almost every time. BLS reports on everyone at work — including the 7.71 million people in food preparation and serving occupations who telework at 0.9% — and gets 22.2%. SWAA reports on full-time wage and salary employees and finds 12% fully remote plus 26% hybrid, or 38% with some flexibility. Employer-panel surveys narrow further, to remote-capable workers only, which excludes most of the on-site economy before the counting starts. None of the three is wrong; they are answering different questions. A remote work statistic quoted without its denominator is close to meaningless, and any figure above 50% is almost certainly drawn from a remote-capable base rather than the whole workforce.
Has the federal return-to-office mandate reduced federal telework to zero?
No. In July 2026, 16.4% of federal government workers — 601,000 of the 3,672,000 at work — still teleworked or worked at home for pay in the reference week. The January 2025 memorandum explicitly allowed agency heads to grant exemptions, and GAO documented that at the Department of Defense alone, 62,000 employees had not returned as of July 2025, 17,000 of them under reasonable accommodations.
Which occupation has the best odds of finding remote work in 2026?
Computer and mathematical occupations, at a 62.5% telework rate with 36.1% fully remote — the highest fully remote share of any occupation group BLS publishes. Business and financial operations follows at 53.9%, and legal at 50.8%, though legal skews heavily hybrid rather than fully remote. If you are targeting fully remote specifically, the fully-remote column matters more than the headline telework rate.
Is the return-to-office push actually raising office attendance, or just changing job titles?
The BLS on-site share is the cleanest available answer, and it has moved very little: 77.2% in 2024, 77.6% in 2025, 77.8% in July 2026. Across two annual averages that is a 0.4-point shift. What moved measurably was the composition of the remaining telework — the Axis hybrid conversion rate fell from 53.1% to 50.5%, meaning a larger share of teleworkers are now fully remote rather than hybrid.
How does the return to office affect salary negotiation and comp bands?
The data supports one specific argument rather than a general one: location flexibility varies far more by occupation than by employer. A candidate in an occupation band with a 62.5% telework rate is negotiating against an economy-wide norm, not a single company’s policy, and can point to it. A candidate in a band at 24.0% draws far less from market data and far more from the specifics of the role. Axis Intelligence Research does not publish compensation differentials for remote versus on-site work in this dataset, because no primary source currently measures them at occupation-band granularity.
Why is the District of Columbia such an extreme outlier in the state data?
D.C.’s 56.8% telework rate in 2024 was two and a half times the national rate, driven by an employment base concentrated in federal administration, professional services and legal occupations — the three highest-telework categories in the economy. Note the date carefully: 2024 annual averages predate the full implementation of the January 2025 directive, so this reading is a pre-mandate baseline rather than a current picture.
When will state-level 2025 telework data be available?
BLS publishes state telework tables annually through the Local Area Unemployment Statistics program; the 2024 table was last modified September 5, 2025. On that cadence the 2025 table is due around September 2026. It will be the first state-level reading to capture a full year under the federal in-person directive, and the District of Columbia, Maryland and Virginia figures are the ones to watch.
Are hybrid workers actually in the office three days a week?
Close to it, measured two ways. BLS finds that workers who teleworked some but not all hours averaged 17.3 hours per week at home in 2025 — a little over two days of a standard week at home, so roughly three in the office. SWAA finds employer plans clustering at 1.3 to 1.5 work-from-home days per week, which implies three and a half to four office days. The gap between the two reflects different populations and different questions, and the honest range is two to four office days depending on which you weight.
Can I republish these charts and figures?
Yes. Everything in this dataset is licensed CC BY 4.0, including the Axis Office Reabsorption Index readings and the state rankings. Attribute to Axis Intelligence Research and link back where practical. The underlying BLS data is U.S. government work and is in the public domain; the index construction, the state rankings and the analysis are ours.