Remote Work Data: What 907M Job Postings Actually Show

Looking at the Actual Data

The narrative around remote work has gone through at least three distinct phases since 2020. First, it was an emergency response. Then it was the future of work. Now, depending on which headline you read, it is either firmly established or in full retreat.

We looked at the actual numbers. Across 907.5 million job postings spanning 2022 through early 2025, drawn from Indeed, LinkedIn, 200,000+ employer ATS career portals, and dozens of additional sources, here is what the work mode classifications show. For details on how we structure these across our 82 fields, see the data schema.

The Year-by-Year Picture

Three patterns stand out.

Remote postings peaked in 2022, declined through 2024, then rebounded in 2025. The 6.3 million remote postings in 2022 dropped to 4.9 million in 2024 before jumping to 9.5 million in 2025. The 2025 number needs context: total posting volume also increased significantly (from 214.6M to 272.9M), so the remote rate returned to its 2022 level (3.5%) rather than exceeding it. The return-to-office narrative holds for 2023 and 2024, but 2025 suggests stabilization rather than continued decline.

Hybrid is the quiet winner, more than doubling from 2022 to 2025. Hybrid postings grew from 1.6 million in 2022 to 3.9 million in 2025. While hybrid remains a small fraction of total postings (1.4% in 2025), its consistent growth in both absolute and percentage terms tells a different story than the binary "remote vs. office" framing that dominates public discourse.

On-site remains the overwhelming majority at a 25-to-1 ratio. Even at remote work's 2022 peak, on-site postings outnumbered remote postings by more than 22 to 1. In 2025, the ratio is approximately 25 to 1. Healthcare, construction, retail, hospitality, manufacturing, and logistics collectively represent the majority of job postings, and these industries are inherently location-dependent.

The Source Gap: ATS Feeds vs. Job Boards

One of the most revealing findings is how dramatically the remote rate varies by data source.

Greenhouse and Lever show remote rates 5 to 7 times higher than Indeed or SimplyHired. Both are ATS platforms used predominantly by technology companies and venture-backed startups. Indeed and SimplyHired serve the full breadth of the labor market including healthcare, retail, and blue-collar occupations.

This gap is not primarily a data quality issue. It reflects a genuine compositional difference: the types of companies that use Greenhouse and Lever are disproportionately technology companies, and technology companies are disproportionately likely to offer remote work. When an analyst reports that "20% of jobs are remote" based on ATS feed data, or "3% of jobs are remote" based on Indeed data, both numbers are correct for their respective source populations. Neither describes the labor market as a whole.

This is why multi-source datasets matter. Our dataset combines all of these sources and deduplicates across them, producing a blended remote rate of approximately 3% to 4% for 2023 and beyond. For details on how we handle deduplication and source blending, see our methodology. You can also compare source-level field coverage on the providers page.

Remote by Seniority

The relationship between seniority and remote availability is consistent, with directors more than twice as likely to have remote postings as entry-level workers.

• Director: 6.8% remote rate
• Associate: 5.8%
• Internship: 4.3%
• Mid-Senior: 3.8%
• Executive: 3.4%
• Entry Level: 3.0%

The "Associate" level (5.8%) being notably higher than "Mid-Senior" (3.8%) likely reflects the prevalence of remote-friendly associate roles in finance, consulting, and technology. The relatively high internship rate (4.3%) may reflect post-pandemic normalization of remote internships in technology and professional services.

The executive level (3.4%) being lower than director level (6.8%) is initially counterintuitive. The likely explanation is that executive roles carry board-facing, client-facing, and leadership responsibilities that require physical presence, even when the same organization offers remote work to director-level individual contributors. Our seniority classifications come from the same NLP enrichment pipeline described in the model overview, which classifies 100% of postings into a seniority tier.

The Surprising Top Remote Job Titles

Ask most people to name the most common remote job, and they will say software engineer. The data says otherwise. The top remote titles by posting volume (2023 onward):

Tutoring dominates remote work by volume, not software engineering. Math Tutor and Tutor combined account for 451,000 remote postings, nearly five times the volume of the next category. This reflects the massive growth of online tutoring platforms in the post-pandemic education market.

Software engineering, despite its cultural association with remote work, does not appear in the top ten by volume because the absolute number of software engineering postings is smaller than the number of tutoring, sales, and customer service postings. Remote work is not primarily a technology phenomenon. It is an economic phenomenon that spans every industry where the work product is digital. For a complementary view of what occupations look like from a skills perspective, see our post on skills extraction across 907M records.

The Structural Data Break

Understanding remote work data requires understanding a fundamental structural break in the underlying fields.

Before 2020, "remote work" as a structured field essentially did not exist in job posting data. The concept existed, but it lived in free-text descriptions, not in machine-readable metadata. By 2021, LinkedIn, Indeed, and most major ATS platforms had added work mode as a structured field. By 2023, the field had become standard with classification coverage approaching 100%.

Any time-series analysis of remote work trends must account for this data availability boundary. Comparing 2019 remote rates to 2023 remote rates is comparing absence of data to presence of data, not measuring a trend. Meaningful trend analysis begins in 2022, when structured work mode data became reliably available across sources. We cover how field availability changes over time in our glossary, which includes definitions for work mode, hybrid, and related classifications.

Connecting Remote Trends to Geography and Salary

Remote work data does not exist in a vacuum. It intersects with two other dimensions that matter for workforce planning.

First, geography. A "remote" posting originating in San Francisco has different salary implications than one originating in Austin or Omaha. The geographic distribution of remote postings is heavily skewed toward coastal metro areas and tech hubs, which means the salary data associated with remote roles skews higher than the national average for the same occupation. Our geography data tracks parsed locations across all postings, including remote-tagged ones.

Second, compensation. The doubling of hybrid postings from 1.6 million (2022) to 3.9 million (2025) complicates compensation benchmarking directly. A hybrid role in New York may command different pay than a fully remote or on-site role in the same city. The salary transparency post covers how state-level disclosure laws have changed salary data availability over this same 2022 to 2025 period.

What the Numbers Add Up To

Across 907.5 million job postings, the remote rate has stabilized around 3% to 4%. Hybrid is growing faster than remote in percentage terms. On-site remains dominant by a factor of 25 to 1. And the most common remote jobs are not software engineers. They are tutors.

These numbers reflect the full breadth of the labor market as captured by multi-source data, not the technology-sector slice that dominates media coverage. For workforce planners, compensation analysts, and HR technology builders, the distinction matters.

If you want to explore remote, hybrid, and on-site classifications across your target market, request a sample.