Workflow Automation Statistics 2026–2027
By Axis Intelligence Research
Co-author: Elena Rodriguez | Last updated: November 14, 2026 | License: CC BY 4.0
Key takeaways
- 5.35% of EU enterprises with 10+ employees used AI-based workflow automation or software robots in 2025, against 46.45% that run an ERP package (Eurostat). Most firms own the system of record; few have the automation layer on top.
- Size decides adoption: 24.42% of large EU enterprises use the layer, against 4.13% of small ones, a 5.91x gap.
- In the US, 19.8% of firms reported using AI in the two weeks before 3 May 2026 (U.S. Census Bureau). In the cycle ending 1 November 2026, 44.1% of firms with 250+ employees used AI, against 22.8% of firms with 1–4 employees (Census BTOS).
- The payoff is real but concentrated: 37% of surveyed companies attribute some EBIT impact to AI, and only about 6% are high performers (McKinsey).
- A $15 monthly licence pays back in about 39.5 minutes of office-administration work per month at US median pay (Axis calculation).
- Open source leads developer traction: n8n holds 210,443 GitHub stars, 4.43x Apache Airflow.
How many companies use workflow automation?
Workflow automation is not one statistic, because the question changes with the source. Three measurements matter, and they answer different things.
| Measure | Value | Population | Source |
|---|---|---|---|
| AI workflow automation, decision support or software RPA | 5.35% | EU enterprises, 10+ employees, 2025 | Eurostat |
| Any AI technology | 19.95% | EU enterprises, 10+ employees, 2025 | Eurostat |
| ERP, CRM or BI use | 53% | EU enterprises, 10+ employees, 2025 | Eurostat news |
| AI in a business function | 18% of firms; 32% of employment | US employer firms, Nov 2025–Jan 2026 | Census CES-26-25 |
| AI used in the last two weeks | 19.8% | US firms, to 3 May 2026 | Census BTOS |
| AI used in the last two weeks, 250+ employees | 44.1% | US firms, cycle ending 1 Nov 2026 | Census BTOS |
| AI used in the last two weeks, 1–4 employees | 22.8% | US firms, cycle ending 1 Nov 2026 | Census BTOS |
The first row is the closest official proxy for “workflow automation”. Eurostat groups three things in one indicator: workflow automation driven by AI, decision support, and software RPA. The EU-27 reading is 5.35%, so roughly one enterprise in nineteen.
The employment-weighted US figure explains why vendors sound more confident than the firm counts suggest. Large employers adopt first, so 32% of employment sits in AI-using firms while 18% of firms use AI.
Which countries lead on workflow automation?
Denmark leads the EU-27 at 15.19%, followed by Finland at 15.03% and Belgium at 10.84%. Greece reports 1.89% and Romania 2.18%. The top-to-bottom spread is 8.04x.
Sweden (8.79%), Lithuania (8.69%) and the Netherlands (8.45%) follow the leaders. Poland sits at 2.60%, Italy at 2.86% and France at 4.48%. A total of 15 of the 27 member states sit below the EU aggregate of 5.35%, because the aggregate is weighted toward large economies with bigger enterprise bases.
On broader AI use the same ordering holds: Denmark reaches 42.03% of enterprises and Romania 5.21%.
Where the stack is deep: the Workflow Stack Depth Index (WSDI)
A single adoption rate hides whether a country is building on a base. Axis Intelligence Research built the Workflow Stack Depth Index (WSDI) to read the stack as layers: an automation layer on top of the two systems of record, ERP and CRM.
WSDI = 50% × (workflow layer ÷ 15.19) + 25% × (ERP ÷ 66.25) + 25% × (CRM ÷ 55.79), each × 100. Each divisor is the best EU-27 country reading in the 2025 survey. A score of 100 would mean leading in all three. Inputs are Eurostat’s E_AI_TPA, E_ERP1 and E_CRM1 indicators for enterprises with 10+ employees. All 27 country scores and every component are in the CSV.
| Rank | Country | WSDI |
|---|---|---|
| 1 | Finland | 96.4 |
| 2 | Denmark | 94.0 |
| 3 | Belgium | 77.8 |
| 4 | Netherlands | 71.9 |
| 5 | Spain | 59.7 |
| 6 | Malta | 58.5 |
| 7 | Luxembourg | 58.1 |
| 8 | Lithuania | 57.6 |
| 9 | Sweden | 54.7 |
| 10 | Germany | 52.6 |
| … | EU-27 aggregate | 47.9 |
| 25 | Bulgaria | 27.8 |
| 26 | Croatia | 26.4 |
| 27 | Romania | 24.5 |
Two findings stand out:
- Finland edges Denmark once CRM depth counts, even though Denmark leads on the automation layer alone.
- Five countries have a strong ERP base but a thin automation layer: Czechia, Greece, France, Italy and Portugal. Their ERP use is above the EU average of 46.45% while their workflow layer sits below 5.35%. These are the markets where a connector or RPA project has an installed base waiting.
The ranking is stable under other weightings: a Spearman rank correlation of 0.982 between the 50/25/25 weights and equal thirds, with the same top five. Sweden’s ERP input carries a Eurostat break-in-series flag, so read its rank with care.
Workflow automation by company size
| EU-27 enterprise size | AI workflow automation or RPA | Any AI | ERP | CRM |
|---|---|---|---|---|
| Small (10–49 employees) | 4.13% | 17% | 41% | 25% |
| Medium (50–249) | 8.67% | 30.36% | n/a | n/a |
| Large (250+) | 24.42% | 55.03% | 89% | 65% |
A large EU enterprise is 5.91x as likely as a small one to run the automation layer. Small firms are not missing ERP so much as the layer above it: ERP sits at 41% for small firms, yet only 4.13% automate workflows.
In the US the Census Bureau shows the same pattern. In the cycle ending 1 November 2026, AI use rises with every size class: 22.8% of firms with 1–4 employees, 29.2% at 20–49, 40.0% at 100–249 and 44.1% at 250+, a 1.93× gap between the largest and smallest class. Employee use of generative AI follows the same ladder, at 55.1% of 250+ firms and 24.5% of 1–4 firms. In the May 2026 reading, information led by sector at 39.7% and finance and insurance followed at 33.9%.
Which US states lead? In the same November cycle, Wyoming (40.1%) leads and New Hampshire (12.8%) trails among the 51 areas with a published estimate, a 27.3-point spread; the median state reads 23.3%. State readings are noisy: Wyoming stood at 14.4% one cycle earlier, so Axis ranks states on a six-cycle mean in the dashboard. Census suppresses some estimates, for example Alaska in the latest cycle.
What do companies actually automate?
Among EU enterprises that use AI, 34.70% do so for marketing or sales and 31.05% for business administration or management. Logistics trails at 6.08%.
US adopters tell a similar story. The Census working paper shows 52% of adopting firms use AI in sales and marketing, 45% in strategy and business development, and 41% in IT.
Depth is shallow. 65% of US firms limit AI to three or fewer tasks, 57% of AI users integrate it into three or fewer functions, and 66% use it only to augment tasks rather than replace them. Only 2% of firms reported AI-related employment decreases. Another 22% of firms expect to start within six months.
Does workflow automation pay off?
McKinsey’s 2026 survey (1,719 respondents, 97 countries, fielded 4 May to 8 June 2026) gives the clearest read on payoff:
- 44% of respondents are scaling AI enterprise-wide, up from 38% a year earlier.
- 37% attribute some EBIT impact to AI, essentially unchanged from 2025.
- About 6% qualify as high performers, defined as attributing at least 5% of EBIT to AI.
- High performers are the ones redesigning work: 55% fundamentally redesigned workflows in the prior-year survey, and McKinsey now reports nearly three-quarters, against about one-quarter of other respondents.
The pattern points to redesign rather than tooling. Companies that automate an unchanged process see less than those that rebuild it.
Gartner’s forecast adds a caution: over 40% of agentic AI projects will be canceled by the end of 2027, while at least 15% of day-to-day work decisions will be made autonomously by 2028 and 33% of enterprise applications will include agentic AI (Gartner). These are predictions, not measurements.
What does workflow automation cost, and when does it break even?
List prices are low relative to labor. Microsoft lists Power Automate Premium at $15 per user per month, and Process and Hosted Process plans at $150 and $215 per bot per month (Microsoft). Zapier lists a free plan with 100 tasks a month and paid plans starting from $19.99 (Professional) and $69 (Team) (Zapier).
Axis converts these into break-even time using BLS median pay (BLS): minutes = list price ÷ (annual median pay ÷ 2,080 hours) × 60. This counts wages only, not build time, maintenance or error cost.
| US occupation (median pay, May 2025) | Power Automate Premium ($15) | Zapier Professional (from $19.99) |
|---|---|---|
| Office and administrative support ($47,450) | 39.5 min/month | 52.6 min/month |
| Bookkeeping, accounting and auditing clerks ($50,670) | 36.9 min | 49.2 min |
| Customer service representatives ($44,770) | 41.8 min | 55.7 min |
| Secretaries and administrative assistants ($48,310) | 38.7 min | 51.6 min |
| General office clerks ($45,010) | 41.6 min | 55.4 min |
| Receptionists ($38,010) | 49.3 min | 65.6 min |
A $150 Power Automate Process bot breaks even at 6.6 hours a month against the office-administration median and 6.2 hours against bookkeeping clerks. For context, BLS counts 1,532,400 bookkeeping clerk jobs and projects 85,600 fewer of them over 2025–2035 (a 6% decline) in its bookkeeping profile.
Which workflow automation tools are gaining traction?
Open-source repositories give a public, dated measure of developer interest. In the 14 November 2026 GitHub snapshot (workflow-automation topic):
| Repository | Stars | Forks | Forks per 100 stars |
|---|---|---|---|
| n8n-io/n8n | 210,443 | 61,498 | 29.2 |
| FlowiseAI/Flowise | 55,570 | 25,094 | 45.2 |
| apache/airflow | 47,474 | 18,358 | 38.7 |
| ToolJet/ToolJet | 41,229 | 5,508 | 13.4 |
| conductor-oss/conductor | 32,470 | 1,069 | 3.3 |
| Budibase/budibase | 28,414 | 2,238 | 7.9 |
| activepieces/activepieces | 25,519 | 4,461 | 17.5 |
n8n holds 4.43x the stars of Airflow and 8.25x those of Activepieces. Since its June 2019 creation it has averaged 77.9 new stars per day. Forks per 100 stars is a rougher signal of builders rather than spectators: Flowise (45.2) and Airflow (38.7) sit well above Activepieces (17.5). Stars measure attention, not production use.
On the commercial side, UiPath reported annualized renewal run-rate of $1,938 million for the quarter ended 31 July 2026, up 12% year over year, with revenue of $410.256 million (+13%) and dollar-based net retention of 109% (UiPath). Zapier says it serves 3.4 million companies, that 69% of the Fortune 1000 use it and that 81 billion tasks have been automated, with 9,000+ apps (Zapier); these are company-reported and undated.
For market size, Mordor Intelligence estimates $26.01 billion in 2026, rising to $40.77 billion by 2031 at a 9.41% CAGR (Mordor Intelligence). UiPath’s ARR equals about 7.45% of that 2026 estimate, a scale check rather than a market share, since the two measures are defined differently.
Workflow automation statistics 2026 vs 2027: what changes next
The 2025 Eurostat survey is the latest official baseline. A 2026 reading would let Axis publish a second WSDI reading and show which countries moved. Until then, this page is refreshed in place so the URL keeps its history. Watch three things:
- Depth, not breadth. With 65% of US firms keeping AI to three or fewer tasks, growth is likely to come from adding functions inside firms that already use it.
- Agent projects. Gartner’s 40% cancellation call is the number to test against next year’s surveys.
- Mid-sized firms. The step from 4.13% (small) to 8.67% (medium) to 24.42% (large) shows where the next adopters are.
Frequently asked questions
What percentage of companies use workflow automation?
In the EU, 5.35% of enterprises with 10+ employees used AI workflow automation, decision support or software RPA in 2025 (Eurostat). In the US, 19.8% of firms reported using AI in the two weeks to 3 May 2026 (Census Bureau). The two numbers measure different things and should not be compared directly.
Which country uses workflow automation the most?
Denmark leads on the automation layer alone at 15.19% of enterprises, with Finland at 15.03%. On the Axis WSDI, which also counts ERP and CRM depth, Finland ranks first at 96.4 and Denmark second at 94.0.
How many hours of work does workflow automation need to save to pay for itself?
At US median pay for office and administrative support, a $15 per month Power Automate Premium licence breaks even at about 39.5 minutes a month. A $150 per month bot breaks even at about 6.6 hours. These figures count wages only.
Do small businesses use workflow automation?
Less than large ones: 4.13% of small EU enterprises use the layer, against 8.67% of medium and 24.42% of large enterprises. Small firms are closer on ERP (41%) than on automation.
Is workflow automation actually profitable?
For many firms, not measurably yet. McKinsey finds 37% of respondents attribute some EBIT impact to AI, and about 6% attribute 5% or more. High performers are far more likely to have redesigned workflows.
What is the most popular open-source workflow automation tool?
By GitHub stars on 14 November 2026, n8n leads the workflow-automation topic with 210,443 stars, ahead of Flowise (55,570) and Apache Airflow (47,474).
What do companies automate first?
Among EU enterprises using AI, marketing or sales (34.70%) and business administration or management (31.05%) lead; logistics is at 6.08%. In US adopting firms, sales and marketing leads at 52%.
What is the Workflow Stack Depth Index?
WSDI is an Axis Intelligence Research composite of a country’s workflow-automation layer (50%), ERP use (25%) and CRM use (25%), each scaled to the best EU-27 reading and multiplied by 100. The EU-27 aggregate reads 47.9.
Methodology and sources
All figures come from the sources linked in the text, retrieved on 14 November 2026. Every number in this article has a row in workflow-automation-statistics.csv, with its source, URL, qualifier and retrieval date. Figures calculated by Axis carry axis_calculated = yes and a stated formula.
- Eurostat: datasets
isoc_eb_ai(E_AI_TPA) andisoc_eb_iip(E_ERP1, E_CRM1), 2025 survey, enterprises with 10+ employees excluding the financial sector; plus Statistics Explained. - U.S. Census Bureau: BTOS data downloads (state and employment size class files), Business Trends and Outlook Survey story and CES working paper 26-25.
- McKinsey & Company: The state of AI in 2026.
- Gartner: agentic AI prediction press release.
- U.S. Bureau of Labor Statistics: Office and Administrative Support and bookkeeping clerks.
- Vendors: Microsoft, Zapier pricing, Zapier about, UiPath investor relations. Vendor figures are company-reported.
- GitHub: REST API search snapshot of 14 November 2026; topic page.
- Mordor Intelligence: market size estimate, a research-firm estimate.