Published · Phulax LLC
How much work can companies with 10 to 200 employees automate?
For the average employee of a U.S. company with 10 to 199 employees, about 15-20% of the work week is admin and data work that current tools can technically do: entering data, processing invoices and documents, approvals, compliance records, scheduling and reports. That is roughly 6-8 hours a week, depending on the industry.
If a first project removes 30% of that, the average company in this size band gets back 40-93 staff hours a week, worth about $30k-$230k a year in payroll. Across all 1,307,401 U.S. companies of this size, that adds up to about 4.1 billion hours and $110 billion a year.
Who this covers
- 1.31MU.S. companies with 10-199 employees
- 39.1Mpeople working at them
- 27,192of those companies in Wisconsin, with 804,423 employees. The most common: accommodation and food services (4,804), health care and social assistance (3,613), manufacturing (3,097), retail trade (2,803), construction (2,335).
Source: U.S. Census Bureau, Statistics of U.S. Businesses 2022, by enterprise size.
By industry: the average company with 10-199 employees
“Automatable share” is the technical ceiling. Hours and dollars are the planning case: a first project that removes 30% of the ceiling. Sorted by dollars per company.
| Industry | U.S. companies | Wisconsin | Avg. staff | Automatable share | Hours a week | Payroll a year |
|---|---|---|---|---|---|---|
| Finance and insurance | 26,547 | 601 | 34.3 | 19.8% | 81 | $230,138 |
| Utilities | 1,706 | 33 | 43.9 | 17.6% | 93 | $219,842 |
| Information | 16,041 | 279 | 34.9 | 18.3% | 77 | $175,828 |
| Professional, scientific and technical services | 108,257 | 1,650 | 29.2 | 19.3% | 68 | $158,116 |
| Mining, oil and gas | 4,522 | 36 | 33.5 | 16.5% | 66 | $150,810 |
| Wholesale trade | 65,437 | 1,433 | 31.2 | 17.4% | 65 | $123,555 |
| Manufacturing | 88,566 | 3,097 | 37 | 17.3% | 77 | $112,197 |
| Construction | 130,625 | 2,335 | 28.3 | 16.6% | 56 | $100,183 |
| Real estate and rental | 26,569 | 455 | 27.4 | 17.6% | 58 | $96,992 |
| Management of companies | 9,530 | 207 | 17.4 | 19.2% | 40 | $91,431 |
| Transportation and warehousing | 39,888 | 1,070 | 33.2 | 16.9% | 67 | $86,406 |
| Health care and social assistance | 180,549 | 3,613 | 32.5 | 17.9% | 70 | $83,515 |
| Admin, support and waste services | 69,685 | 1,136 | 32.6 | 17% | 66 | $81,875 |
| Educational services | 32,031 | 591 | 36.5 | 17% | 75 | $72,340 |
| Agriculture, forestry and fishing | 2,999 | 28 | 26.3 | 15.8% | 50 | $67,358 |
| Retail trade | 124,583 | 2,803 | 27.1 | 16% | 52 | $60,669 |
| Other services (repair, personal care and more) | 108,333 | 2,294 | 24 | 16.6% | 48 | $48,577 |
| Arts, entertainment and recreation | 31,175 | 724 | 31.6 | 15.7% | 59 | $47,281 |
| Accommodation and food services | 240,192 | 4,804 | 27.6 | 14.5% | 48 | $29,585 |
Payroll is wages only, before benefits and payroll taxes, at each industry’s average pay in companies of this size. Hours a week are for the whole company.
By job: where the hours sit inside a company
The automatable share varies much more by job than by industry. Selected jobs, as a share of a 40-hour week (technical ceiling), from 759 occupations analyzed:
- 27%Billing and Posting Clerks (10.7 hours a week)
- 26%Payroll and Timekeeping Clerks (10.5 hours a week)
- 26%Insurance Claims and Policy Processing Clerks (10.2 hours a week)
- 25%Bookkeeping, Accounting, and Auditing Clerks (9.9 hours a week)
- 24%Secretaries and Administrative Assistants, Except Legal, Medical, and Executive (9.6 hours a week)
- 23%Data Entry Keyers (9.4 hours a week)
- 20%Receptionists and Information Clerks (7.9 hours a week)
- 19%Customer Service Representatives (7.7 hours a week)
- 19%Heavy and Tractor-Trailer Truck Drivers (7.6 hours a week)
- 18%Dispatchers, Except Police, Fire, and Ambulance (7.3 hours a week)
- 18%General and Operations Managers (7.1 hours a week)
- 16%Heating, Air Conditioning, and Refrigeration Mechanics and Installers (6.5 hours a week)
- 14%Construction Laborers (5.6 hours a week)
- 11%Landscaping and Groundskeeping Workers (4.4 hours a week)
- 10%Waiters and Waitresses (4.1 hours a week)
Field and floor jobs score lower per person, but a construction or trucking company has many more of them, so their job notes, time sheets, delivery records and inspection forms often add up to more hours than the office’s.
By type of work: total hours a year across all U.S. companies of this size
| Work that can be automated | Hours a year (planning case) |
|---|---|
| Looking up information * | 782 million |
| Document and invoice processing | 511 million |
| Compliance and certification tracking | 503 million |
| Monitoring and alerts * | 496 million |
| Intake and data entry | 487 million |
| Scheduling and dispatch | 375 million |
| Approvals and admin workflows | 289 million |
| Internal messages and handoffs | 223 million |
| Inventory and spend control | 219 million |
| Customer messages and follow-up | 118 million |
| Reports | 107 million |
* Softer estimates. Almost every job rates “getting information” and “monitoring” as important, including jobs where that means watching a job site, a road or a machine, so software can only assist with part of these hours.
The firmest targets are document and invoice processing, compliance and certification tracking, intake and data entry, scheduling and dispatch, approvals and admin workflows: about 2.2 billion hours a year together. They also have clear before-and-after measures, such as cost per invoice and days to approve, which makes them the natural place to start.
What published studies say
- 32%of U.S. firms with 100-249 employees used AI in the two weeks before the survey, against 19.8% of all firms (period ending May 3, 2026). Census Bureau, Business Trends and Outlook Survey
- 7%of small employer firms that use AI have it fully built into how they work. 46% of firms use AI in some way. Convenience sample of 6,525 firms with 1-499 employees. Federal Reserve Small Business Credit Survey, 2026 report
- $5.83median cost to process one invoice; the best quarter of organizations spend $2.07 or less (1,485 organizations, 2018 data). APQC, reported by CFO.com
- $2.78per invoice and 3.1 days to process for best-in-class companies, against $12.88 and 17.4 days for everyone else. Ardent Partners, State of ePayables 2024
- 14 hrsa week spent chasing customer payments, on average, by the 65% of companies with 25-200 employees that do it (vendor-commissioned survey, 2021). Intuit QuickBooks and Wakefield Research
- 5.4%of work hours saved by people who use generative AI; averaged across all workers it is 1.4%. Federal Reserve Bank of St. Louis
- +14%more customer issues resolved per hour by support agents given an AI assistant, and +34% for the least experienced agents (5,179 agents). National Bureau of Economic Research, working paper 31161
Put together: many small companies have tried AI, few have built it into how work gets done, and broad “AI for everyone” tools save little. The big gains come from systems built for one workflow at a time.
How we worked it out
- Activities per job. O*NET 31.0 rates how important each of 41 work activities is to every job. We turned importance into a share of the week, weighting each job’s core activities most (importance above “not important”, squared).
- What can be automated. We counted eleven admin and data activities and applied McKinsey Global Institute’s published estimates of how much of each activity type current technology can do (2017): 64% of collecting data, 69% of processing it, 20% of communicating with people and 18% of applying expertise. Physical, caring, selling, creative and managing work counts as zero.
- Jobs per industry. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025: how many people do each job in each industry.
- Company size and pay. Census Bureau, Statistics of U.S. Businesses 2022, enterprise sizes 10-14 through 150-199: company counts, employees and payroll.
- Planning case. We assume a first project removes 30% of the technical ceiling, because exceptions, review and adoption always cut into it.
Limits. The job mix is for each industry as a whole, not only small companies, where owners and general staff do more of the admin. O*NET measures importance, not hours; with a flatter conversion the dollar figures drop by 0-8% and the ranking barely changes. The McKinsey rates predate today’s AI tools, which likely raise the ceiling for writing-heavy work. Dollars are wages only. These are national estimates, not measurements of any one company.
Data as of 2026-10-17. All inputs are free public data: U.S. Census Bureau and Bureau of Labor Statistics (public domain); O*NET 31.0 Database by the U.S. Department of Labor, Employment and Training Administration, used under CC BY 4.0 (O*NET is a trademark of USDOL/ETA); McKinsey Global Institute, A Future That Works (2017).
Questions
- How much of a small company's work can be automated?
- For the average employee of a U.S. company with 10 to 199 employees, about 15-20% of the work week is admin and data work that current tools can technically do, roughly 6-8 hours a week depending on the industry. That is a ceiling. A realistic first project removes a share of it; we plan on 30%.
- Which jobs have the most automatable work?
- Back-office jobs. Billing and Posting Clerks are at about 27% of the week (10.7 hours), along with payroll, insurance claims and bookkeeping clerks. Hands-on jobs are lowest, such as waiters and waitresses at about 10%.
- Which industries gain the most?
- Per company, finance and insurance (about $230k a year in payroll), professional services, wholesale, manufacturing and construction, because they combine high pay with a lot of paperwork. Restaurants and hotels gain the least per company.
- Is this the same as how much time AI saves today?
- No. This is what current tools could technically do with admin and data work. Surveys of real use find much smaller savings from general AI chat tools, about 5% of hours for the people who use them. The larger savings come from systems built for one specific workflow, such as invoice processing or scheduling.
- Can I get these numbers for my own company?
- Yes. A staff list with job titles, headcount and pay gives a first estimate by job and by system. The real numbers come from timing your actual workflows, which is what our cost-reduction analysis does.
What would this look like at your company?
These are national averages. We time your real workflows, rank what to automate and what to leave alone, and build the systems that remove the most hours. About the cost-reduction analysis