Workflow automation statistics: 72 figures on business process automation in 2026

The latest workflow automation statistics show a gap between plans and practice. Large organizations say 48% of their processes are automated and 95% credit automation with growth, yet only 27% of their applications are connected and just 11% of agentic AI use cases reached production last year.

The work that stays manual is often the paperwork inside the workflow: invoices, forms and files a person reads before the next system can act. We grouped the figures below by the question they answer. Each one is numbered so you can link to a single statistic, and we checked every one at the publisher's own page or report in September 2026.

Use the spread as a warning. Surveys of large companies about AI in general report far higher numbers than surveys that ask whether a process runs end to end without people.

Popular workflow automation statistics, checked#

Many of the figures repeated across automation round-ups are years old, describe something narrower than the quote, or have no source at all. We traced the most common ones back to where they started.

How many organizations automate workflows#

1. Nearly nine in ten organizations use AI regularly in at least one business function, and 44% say AI is now scaling across the enterprise, up from 38% a year earlier (McKinsey, The state of AI in 2026).

2. Large organizations have automated 48% of their processes on average and believe they could reach 64% (Camunda survey of 1,150 decision makers, 2026).

3. 72% of US enterprises with 1,000 or more employees are using or testing AI agents, and 30% of leaders see the most potential in automating routine workflows (Zapier survey of 525 executives, 2025).

4. 40% of organizations with more than $1 billion in revenue report scaling AI agents in at least one function, up from 27% a year earlier. Among smaller organizations the share stayed flat at 22% (McKinsey, 2026).

5. Only 31% of large enterprises say AI is fully embedded in their business. 35% use it among select teams and 11% are stuck in pilot mode (UiPath survey of 590 executives and IT practitioners, 2026).

6. Just 17% to 20% of US businesses used AI to produce goods or services between December 2025 and May 2026 (US Census Bureau, Business Trends and Outlook Survey).

7. AI use rises with company size. 37% of US firms with 250 or more employees use AI, against under 20% of firms with four or fewer (US Census Bureau, 2026).

8. 33.9% of US finance and insurance businesses use AI, well above the 19.8% national rate (US Census Bureau, 2026).

9. In a January 2025 Gartner webinar poll of 3,412 attendees, 19% said their organization had made significant investments in agentic AI, and 31% were waiting or unsure (Gartner, 2025).

Where automation stalls between systems#

The average organization runs close to a thousand applications, and most of them were never built to pass work to each other. That is the plumbing problem behind most stalled automation projects.

10. The average organization runs 957 applications, and only 27% of them are connected (MuleSoft 2026 Connectivity Benchmark, 1,050 IT leaders).

11. 82% of IT leaders say data integration is one of the biggest challenges their organization faces in using AI (MuleSoft, 2026).

12. IT teams spend 36% of their time building custom integrations between systems and data, and 26% of IT projects were delivered late over the past year (MuleSoft, 2026).

13. 78% of US enterprises say they are struggling to integrate AI with their existing systems (Zapier survey of 532 executives, 2025).

14. Enterprises name data quality and readiness (38%) and integration with existing workflows and systems (37%) as the top hurdles to scaling agentic AI, ahead of governance and compliance at 33% (UiPath, 2026). What poor data quality costs is in our cost of bad data statistics.

15. 48% of organizations say their AI agents work in silos instead of being part of end-to-end business processes (Camunda, 2026).

16. In Deloitte's intelligent automation survey, the top barriers to scaling were difficulty integrating solutions (62%), lack of skills (55%) and an inability to change business processes (52%). The survey is from 2022, and Deloitte hasn't published a newer edition (Deloitte, 2022).

The paperwork inside the workflow#

Most process maps start after the data exists. In accounts payable, lending and insurance, someone first has to get it off an invoice, a statement or a form, and that step is often still manual.

17. Unstructured data such as documents and media files makes up 70% to 90% of organizational data, according to Gartner. Many round-ups quote 80% to 90%, which is not Gartner's range (Gartner, 2026). Our unstructured data statistics trace where the popular 80% figure comes from.

18. 61% of document processing workflows still include paper, and 48% of organizations expect their paper volumes to grow next year (AIIM and Deep Analysis survey of 600 enterprises, 2025).

19. Only 29% of the roughly 300 billion B2B invoices sent worldwide each year are true e-invoices. In North America the share is 36% (billentis report, 2026).

20. Only 57% of suppliers can send invoices electronically, so accounts payable teams still capture a large share of invoices from PDFs and paper (Ardent Partners, State of ePayables 2025). More AP figures are in our accounts payable statistics.

21. The average accounts payable team spends $9.84 and 8.2 days to process one invoice, and 18.4% of invoices hit an exception (Ardent Partners, 2025).

22. Slow invoice approvals and high exception rates tie as the top accounts payable challenges of 2026, each named by 48% of AP leaders (Ardent Partners, The State of AP 2026).

23. Only about 11% of ocean bills of lading were issued electronically as of 2025. The rest still move on paper (Digital Container Shipping Association).

24. Property and casualty underwriters spend more than 41% of their time on administrative work such as data entry and record keeping (Capgemini survey of 201 underwriters, 2024).

25. 55% of US enterprises say the cost of managing their unstructured data is rising, and only 4% say it is falling (AIIM and Deep Analysis survey of 500+ enterprises, 2024).

26. 92% of US enterprises have found at least one process that AI could improve by working on their unstructured content (AIIM and Deep Analysis, 2024).

27. 65% of organizations say their structured data is at least somewhat ready for AI, but only 39% say the same of their unstructured data, such as emails and PDFs (Harvard Business Review Analytic Services survey sponsored by Hyland, 2026).

28. Gartner predicts that through 2028, teams that build their own tooling to tag unstructured data will pay more than 300% more than teams using existing document and records products (Gartner, 2026).

29. By 2027, 40% of IT spending on data management will go to data that mixes structured and unstructured formats, Gartner predicts (Gartner, 2026).

Document automation and IDP#

30. 78% of enterprises in the AIIM and Deep Analysis survey are now operational with AI in their document processing (AIIM, 2025).

31. 66% of new intelligent document processing projects replace an existing system instead of automating a process for the first time (AIIM, 2025).

32. Only 30% of organizations count a smaller headcount as a sign of a successful document processing project, the lowest of seven success measures in the survey (AIIM and Deep Analysis, 2025).

33. Invoice capture and data extraction is the most widely deployed use of AI in accounts payable, at 58% of AP organizations (Ardent Partners, 2026). It is also the step invoice processing software is built to take off people's desks.

34. Best-in-class AP teams process invoices at 79% lower cost and put 1.8 times as many invoices straight through with no manual touch (Ardent Partners, 2025).

35. The most common enterprise use of AI agents is data management. 47% of enterprises use agents for data entry and extraction, and 41% for document analysis and summarization (Zapier, 2025).

36. To fix problems with their generative AI projects, 35% of business leaders turned to document AI, the same share that turned to process intelligence (ABBYY survey of 1,200 senior managers, 2025).

37. 68% of large insurers are prioritizing automated data entry in their generative AI investment across corporate support functions (EY-Parthenon survey of 100 insurers, 2025). For how that works on submissions and claims, see document processing for insurance.

A straight-through rate is the simplest way to see where your own workflow stands. Count the documents a person had to open, not the ones the software touched.

Our take. The figures on stalled pilots and disconnected systems point the same way. Automate one document-heavy step that feeds a real decision, such as invoice approval or a loan file review, and measure it before widening the scope. A workflow that works for one document type is worth more than a platform that half-covers ten.

Time lost to manual and repetitive work#

These are self-reported surveys, and the shares differ because each one asks about a different kind of lost time.

38. Desk workers say they spend 41% of their time on tasks that are low value, repetitive or add little to their core job (Slack Workforce Lab survey of 10,281 desk workers, 2024).

39. Leaders and teams lose 25% of their time just searching for answers (Atlassian State of Teams 2025, 12,000 knowledge workers).

40. Knowledge workers spend 58% of their day on "work about work," such as chasing status, searching for information and switching apps. Asana hasn't published a newer figure since this 2023 edition (Asana Anatomy of Work Index, 9,615 workers).

41. Knowledge workers estimate better processes would save them 4.9 hours a week, more than six working weeks a year (Asana, 2023).

42. 48% of employees and 52% of leaders say their work feels chaotic and fragmented (Microsoft Work Trend Index, 31,000 workers in 31 markets, 2025).

43. Microsoft 365 users are interrupted by a meeting, email or chat every 2 minutes during core hours. The often-quoted 275 interruptions a day describes the 20% of users who receive the most pings, not the average worker (Microsoft, 2025).

44. 85% of knowledge workers use AI at work, but only 29% have built it into how their work flows (Atlassian State of Teams 2026, 12,035 workers).

How much work could be automated#

45. Technology that exists today could, in theory, automate activities that account for about 57% of US work hours (McKinsey Global Institute, 2025).

46. AI agents could perform tasks that take up 44% of US work hours, and robots another 13% (McKinsey Global Institute, 2025).

47. In McKinsey's midpoint scenario, AI agents and robots could generate about $2.9 trillion a year in US economic value by 2030, if organizations redesign whole workflows instead of single tasks (McKinsey Global Institute, 2025).

48. Generative AI combined with other technologies could automate work activities that absorb 60% to 70% of employees' time. This 2023 estimate is still widely quoted, and MGI's 2025 figure for the US is 57% of hours (McKinsey, 2023).

49. Almost 40% of global employment is exposed to AI, and about 60% of jobs in advanced economies. Exposure is not replacement, and the IMF expects AI to complement many of those jobs (IMF, 2024).

AI agents and RPA in workflows#

50. Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027 because of rising costs, unclear business value or weak risk controls (Gartner, 2025).

51. Gartner estimates only about 130 of the thousands of vendors selling agentic AI are real. Many are "agent washing" existing RPA bots, assistants and chatbots (Gartner, 2025).

52. 71% of organizations say they use AI agents, but only 11% of agentic AI use cases reached production in the past year (Camunda, 2026).

53. 40% of enterprise applications will include task-specific AI agents by the end of 2026, up from less than 5% in 2025, Gartner predicts (Gartner, 2025).

54. By 2028, at least 15% of day-to-day work decisions will be made autonomously through agentic AI, up from 0% in 2024, and 33% of enterprise software applications will include agentic AI (Gartner, 2025).

55. By 2029, agentic AI will resolve 80% of common customer service issues without a person, cutting operating costs by 30%, Gartner predicts (Gartner, 2025).

56. In 2022, 74% of organizations in Deloitte's survey were implementing robotic process automation and 50% were implementing OCR, the two most widely adopted intelligent automation tools (Deloitte, 2022).

Results and return on investment#

57. 95% of large organizations say process automation increased business growth over the past year, up from 87%, and 79% plan to raise automation spending by an average of 20% over two years (Camunda, 2026).

58. Only 37% of organizations attribute any EBIT impact to AI, the same as a year earlier, and 6% qualify as AI high performers. Yet 80% of respondents say AI improved their own productivity (McKinsey, 2026).

59. Most organizations see a satisfactory return on a typical AI use case only after two to four years. Just 6% saw payback in under a year (Deloitte survey of 1,854 executives in Europe and the Middle East, 2025).

60. Among organizations already using agentic AI, just 10% say they are getting significant ROI from it (Deloitte, 2025).

61. Organizations that moved past piloting intelligent automation report an average 32% cost reduction, yet more than half of all respondents had never calculated their savings (Deloitte, 2022).

Automation and AI market size#

Market sizes depend on each firm's definition of the market, so compare figures from the same source over time rather than across firms.

62. Worldwide spending on AI is forecast to reach $2.7 trillion in 2026, up 49.5% from 2025 (Gartner, September 2026).

63. Spending on AI agents and assistants will reach about $29.2 billion in 2026 and $65.5 billion in 2027, up from $16.5 billion in 2025, Gartner forecasts (Gartner, 2026).

64. IDC forecasts AI spending will grow 31.9% a year from 2025 to 2029, reaching $1.3 trillion in 2029 (IDC, 2025).

65. The global robotic process automation market will reach $35.84 billion by 2033, growing 29.0% a year from 2026 (Grand View Research).

Automation and jobs#

66. BLS projects US office and administrative support jobs to fall 4.0% from 2025 to 2035, a loss of 752,100 jobs and the most of any major occupational group, as AI-powered automation is built into workflows. Total US employment grows 3.5% over the same decade (BLS, 2026).

67. Data entry keyer jobs are projected to fall 25.5%, from 131,800 to 98,200, one of the fastest declines of any US occupation (BLS, 2026). Our data entry error statistics cover what manual keying costs in the meantime.

68. Word processors and typists lead all occupations with a projected 34.4% decline, while payroll clerks fall 15.9% and file clerks 15.8% (BLS, 2026).

69. Employers expect structural change to create 170 million jobs and displace 92 million between 2025 and 2030, a net gain of 78 million (World Economic Forum, Future of Jobs Report 2025).

70. Clerical and secretarial roles are expected to see the largest absolute job decline to 2030, with data entry clerks among the fastest-declining roles (World Economic Forum, 2025).

71. Employers say 47% of work tasks are done mainly by people today, 22% mainly by technology and 30% by a combination. By 2030 they expect a near-even split among the three (World Economic Forum, 2025).

72. 40% of employers expect to reduce their workforce where AI can automate tasks, while two-thirds plan to hire people with specific AI skills (World Economic Forum, 2025).

How we picked and checked these figures#

Every figure links to the organization that ran the survey or analysis, and we opened each page or report on September 26 and 27, 2026 to confirm the number and its wording. Where a publisher's site blocks automated access, we read an archived copy of the same page and kept the publisher's own link. We left out forecasts whose target date has passed and numbers we could only find on sites that repeat each other.

Most figures are from 2024 to 2026. A few older ones stay because no newer edition exists and writers still quote them, such as Deloitte's 2022 automation survey and Asana's 2023 index, and each is dated in the text. Several sources sell automation or document processing products, including Camunda, UiPath, Zapier, MuleSoft, Microsoft and ABBYY, and Hyland sponsored the Harvard Business Review Analytic Services study. Their numbers come from their own surveys, and samples range from 100 insurers to 31,000 workers, so compare figures from different surveys with care.

Before you automate the next workflow#

Count what a person does by hand today in the workflow you want to fix: how many documents they open, how many fields they key or correct, and how long a case waits for them. That baseline makes any vendor's result easy to check, including ours.

To see how the document step works on your own files, book a demo or start a free trial.

Frequently asked questions#