If you read marketing blogs, somewhere between 78 and 89 percent of small businesses are using AI. If you read the US Census Bureau's Business Trends and Outlook Survey, the figure sat between 17 and 20 percent from December 2025 through May 2026, with a further 20 to 23 percent expecting to be using it within six months.
The gap between those two pictures is not a rounding difference. It is the difference between a survey of a few hundred self-selecting respondents and a federal survey designed to represent the whole business population. One of these numbers is worth planning against.
We spend a lot of our time building automation for small firms, so the low figure is not convenient for us to publish. It is still the accurate one, and it leads to better decisions than the inflated version does.
What does the Census Bureau data actually say?
The Business Trends and Outlook Survey found overall AI use among US businesses hovering between 17 and 20 percent between December 2025 and May 2026, with 20 to 23 percent expecting to use it within the next six months. Adoption rises sharply with size: around 37 percent of businesses with at least 250 employees reported using AI, and 32 percent of those with 100 to 249 employees.
Firms with four or fewer employees came in under 20 percent, with no significant change over that period. By sector, use reached 39.7 percent in Information and 33.9 percent in Finance and Insurance as of early May 2026, while Retail Trade sat near 14 percent.
Why is the gap between surveys so large?
Mostly because of who answers. Voluntary online polls promoted through technology newsletters and vendor mailing lists reach people already interested in the subject, and a business that has never touched AI has little reason to complete a survey about it. Definitions differ too: some polls count any employee using a chatbot once, while the Census question asks whether the business uses AI to produce goods or services.
Neither figure is dishonest, but only one describes the market a firm competes in. If you plan on the assumption that 89 percent of your competitors have automated, you will conclude you are hopelessly behind. If you plan on the real figure, you will notice the opposite, which is that most of your competitors have not started.
Does low adoption make AI a poor bet?
No, it makes it an available one. The commercial value of a capability comes from the gap between what you can do and what your competitors do, and that gap is currently large. In Retail Trade, where roughly one business in seven uses AI, being the one that answers enquiries in ninety seconds at ten at night is a visible advantage rather than table stakes.
The larger risk in 2026 is not being late. It is buying tools for a problem you have not defined, which is how most small-business automation spending gets wasted. We wrote about where the practical wins land in our piece on how AI automation is saving small businesses hours every month.
Where does automation reliably pay for a small firm?
In the repetitive work that does not need judgement. First response to enquiries, so nobody waits overnight. Routing, so the serious enquiries reach a person and the rest do not consume one. Drafting, where a first version is produced in seconds and a human edits it. Follow-up, which is the single most commonly skipped revenue activity in small businesses because it is boring and nobody owns it.
What does not pay is automating a process that was already broken, or buying a platform to solve a problem that turned out to be one unwritten rule. The honest test before spending anything is whether you can name the task, say how many hours a week it consumes, and say who currently does it. If any of those three answers is vague, the tool will not fix it.
Why does firm size predict adoption so strongly?
Because implementation costs are mostly fixed rather than proportional. Working out what to automate, configuring it and maintaining it takes roughly the same effort whether you have 12 staff or 250, but a large firm spreads that cost across far more transactions and usually has someone whose job includes it. A five-person business is asking an owner to do it in the evenings.
That is the real barrier, and it explains the sub-20 percent figure for the smallest firms better than any story about willingness. It also points at the practical answer for a small firm, which is to automate one narrow, high-frequency task properly rather than attempting a general transformation. One workflow that reliably runs beats five that half work.
How does this connect to marketing?
Directly, because the highest-value automation in a small business usually sits at the point where interest becomes a conversation. Traffic is worth nothing if enquiries wait until morning, and the firms that see real returns are almost always the ones that fixed response and follow-up before buying anything more sophisticated.
That is also why automation and visibility are the same project rather than two. Generating more enquiries into a process that drops them makes the problem worse. Our framework for sequencing this is in the complete guide to digital marketing for accounting firms, and the delivery side sits under AI automation and website design and development.
How do you tell a real saving from a felt one?
By measuring before you buy, which almost nobody does. Automation produces a strong sense of progress because the work looks modern and the output arrives quickly, and that feeling is not evidence. A firm can spend six months and a meaningful budget on tools while the total hours spent on the underlying task stay exactly where they were, and without a baseline nobody notices.
The check is unglamorous. Pick the task, count the hours it consumed last month, and write the number down somewhere you will find it. Then automate, wait six weeks, and count again. If the hours moved, keep going and pick the next task. If they did not, the tool was solving a problem you did not have, and the correct response is to stop paying for it rather than to add a second tool on top.
There is a second measure worth keeping, which is how often the automation is wrong and who catches it. A process that saves four hours a week and quietly sends two incorrect replies is not a saving. It is a liability with good ergonomics, and small firms discover this late because the errors go to people who simply do not reply rather than to people who complain.
What about the sectors that are ahead?
Finance and Insurance at 33.9 percent and Information at 39.7 percent are roughly twice the national rate, which tells you something useful about where client expectations are being set. A CPA firm competing against practices in that 33.9 percent is not competing on whether AI is used, but on whether it is used well. We listed the tools doing most of that work in our roundup of free AI tools US accountants are using in 2026.
For everyone else the read is more forgiving. If your sector sits nearer the national rate, you have time, and you have the advantage of learning from industries that already made the expensive mistakes. The full picture for firms in this market sits on our page for digital marketing for CPA firms in the USA, and the ranking work under search engine optimization.
Does the six-month intention figure mean a wave is coming?
Treat it carefully. Between 20 and 23 percent of businesses said they expected to be using AI within six months, but stated intention has consistently run ahead of actual adoption in this survey, and the headline use rate barely moved across the six months to May 2026 despite similar expectations being recorded earlier.
The reasonable reading is that adoption is grinding upward rather than about to jump. For a firm deciding whether to act now, that is good news twice over: the advantage of moving early lasts longer than a hype cycle would suggest, and there is no need to rush a decision badly in order to avoid being left behind next quarter.
What to do with this
Stop benchmarking against numbers from vendor surveys. Pick the single task in your business that happens most often and needs the least judgement, measure how many hours it takes this month, and automate that one thing. Check the hours again in six weeks. If they did not move, the problem was never the tool.
Triomatic Marketing builds this for small and mid-sized firms across the USA, UK and Pakistan. We are AI-powered and founder-led, and we would rather publish the number that is true than the one that sells more automation. To talk it through, message Aria on WhatsApp via triomaticmarketing.com, or book a free 15-minute discovery call at https://calendly.com/hello-triomaticmarketing/15min.
Frequently asked questions
FAQs
What percentage of US businesses actually use AI in 2026?
The US Census Bureau's Business Trends and Outlook Survey found overall AI use between 17 and 20 percent from December 2025 through May 2026, with a further 20 to 23 percent expecting to adopt within six months. That is far below figures commonly quoted in vendor surveys.
Why do other surveys report much higher AI adoption?
Mostly sampling and definitions. Voluntary online polls promoted through technology newsletters reach people already interested in AI, and some count any single employee using a chatbot, while the Census question asks whether the business uses AI to produce goods or services.
Do small businesses use AI less than large ones?
Substantially less. Around 37 percent of businesses with at least 250 employees reported using AI and 32 percent of those with 100 to 249 employees, while firms with four or fewer employees came in under 20 percent with no significant change over the period.
Which sectors are furthest ahead on AI?
As of early May 2026 the Information sector reached 39.7 percent and Finance and Insurance 33.9 percent, both roughly double the national rate. Retail Trade sat near 14 percent, below the overall figure.
Where does automation actually save a small business time?
In high-frequency work that needs no judgement: first response to enquiries, routing so serious leads reach a person, drafting a first version for a human to edit, and follow-up. Automating a process that is already broken does not help.
How does Triomatic Marketing approach automation?
We start with one narrow, high-frequency task, measure the hours it consumes, and automate that before adding anything else. Usually it is the gap between an enquiry arriving and somebody responding. Book a free 15-minute call at https://calendly.com/hello-triomaticmarketing/15min.