01

Executive summary

Adoption among small firms is broad but shallow, and the smallest local-service firms trail the most.

  • Measured strictly, adoption is about 20%. The Census Bureau's Business Trends and Outlook Survey (BTOS) put it at 19.8% of businesses in early May 2026. Owner surveys that count any use, including a chatbot for drafting emails, report 46% (Federal Reserve) to 76% (Goldman Sachs).
  • Size and sector matter more than region. Firms with 250+ employees reach 37%, while firms with fewer than 20 employees were flat from December 2025 to May 2026. Food service, construction, transportation and agriculture sit below 10%.
  • Most use is light. Of small firms using AI, about half are experimenting and 7% have fully integrated it (Federal Reserve). Goldman Sachs' program alumni put full integration at 14%.
  • Writing, marketing and productivity lead. Sales and marketing is the most common business function among adopting firms (52%).
  • Benefits are reported, job losses are rare. 71% of AI-using small firms report higher productivity, and only 2% of firms report AI-related employment decreases.
  • Barriers differ by group. Non-users mostly say AI does not apply to them. Users cite accuracy, fitting tools to the business, and a lack of training.
02

How many local businesses use AI?

Surveys report anywhere from about 20% to over 75%, and the spread comes from sampling and definitions rather than from real disagreement. The most defensible baseline for all employer businesses is the Census Bureau's 19.8%; owner surveys that count any use, by anyone, run much higher.

SurveyShare using AIWho was askedWhat counts as "use"
Census BTOS19.8% (May 3, 2026)Nationally representative biweekly sample of US businessesAI used in any business function in the past two weeks
Federal Reserve SBCS46% (fall 2025)6,525 small employer firms with 1 to 499 employees; convenience sampleThe business or its employees currently use AI, including AI features of software already in use
BizBuySell63% (Q1 2026)Business owners and entrepreneurs surveyed by a business-for-sale marketplaceSelf-reported AI use
Goldman Sachs 10,000 Small Businesses Voices76% (Jan to Feb 2026)1,256 alumni of Goldman Sachs' small business programCurrently using AI
Intuit QuickBooksAbout 77%34,000+ owners in the US, Canada, the UK and Australia"Regularly" use AI; headline figure as reported by secondary coverage, not confirmed on Intuit's landing page

Three things explain the gap:

  1. Sampling. Census draws a probability sample. The others rely on convenience samples, program alumni or customers, which over-represent engaged owners.
  2. Definition. Counting a free chatbot used for emails yields a far higher rate than counting AI used in production. When the Census Bureau widened its question in November 2025 from "producing goods or services" to "any business function," measured use roughly doubled, from about 10% to about 20%.
  3. Population. Census covers firms of all sizes, while the owner surveys reach small firms through lenders, vendors and program networks.

For a typical local business with fewer than 20 employees, the Census data suggest the real figure is below 20%, with a wide gap between "has tried a chatbot" and "uses AI in operations."

03

Who is adopting

Firm size and industry predict adoption better than anything else in the data; the smallest, most local firms are the slowest. In the Census Bureau's BTOS, AI use rose between December 2025 and May 2026 among firms with 20+ employees but did not change significantly among firms with fewer than 20.

SegmentShare using AIPeriod
All US businesses19.8%Early May 2026
250+ employees37%Period ending May 3, 2026
100 to 249 employees32%Period ending May 3, 2026
4 or fewer employeesUnder 20%Dec 2025 to May 2026
Information sector39.7%Early May 2026
Finance and insurance33.9%Early May 2026
Retail tradeAbout 14%Early May 2026

The sectors most associated with local, storefront and field-service work sit at the bottom. The Minneapolis Fed reports that fewer than 10% of businesses in accommodation and food services, construction, transportation and agriculture use AI. The Census Bureau's earlier BTOS work found the most common reason for not adopting was that AI did not apply to the business.

Geography. Regional evidence is thin. The Philadelphia Fed found that about a third of small firms in the Third District (DE, NJ, PA) use AI, versus 46% nationally. The Philadelphia Fed suggests the gap may reflect industry mix, so it is not clear that location itself drives adoption.

04

What local businesses use AI for

Most small firms use general-purpose assistants for writing, marketing and everyday admin, not custom or industry-specific systems. The Federal Reserve's 2025 Small Business Credit Survey (SBCS) found the most common uses were writing or marketing, individual productivity, and planning or analysis.

  • By business function (Census). Among adopting firms, sales and marketing leads at 52%, followed by strategy and business development (45%) and IT (41%). Within tasks, writing, document analysis and information search dominate generative AI use.
  • By owner reason (BizBuySell, Q1 2026). Owners cite productivity (78%), analysis and insights (60%) and automating routine tasks (56%) as their main reasons. This survey polled business owners and entrepreneurs, so it leans toward people active in the business-for-sale market.
  • By tool (Bluevine, April 2026). 57% of AI-using small firms use ChatGPT and 56% use Google Gemini; 39% use AI for data analysis and 37% for marketing and sales. Most popular tools are free, and 28% spend $25 to $99 a month. This is a lender-sponsored poll fielded over three days.
  • Local view (Intermax, Inland Empire). A regional IT provider's survey of its own market found customer-service automation, content and marketing support, and internal productivity tools as the usual starting points.
05

Depth of adoption

Few small firms have built AI into how they operate. Of small employer firms using AI in the Federal Reserve's survey, about half were experimenting, 44% had partially integrated it into processes, and 7% had fully integrated it. Goldman Sachs' 10,000 Small Businesses Voices survey found 14% fully embedded in core operations, among program alumni rather than a random sample.

The Census Bureau's working paper agrees: 57% of adopting firms use AI in three or fewer business functions, and 65% of firms limit AI to three or fewer tasks.

Momentum is still positive. The Fed survey found another 15% of small firms planned to start within 12 months, and Bluevine reports that owners with no plans to use AI fell from 30% in 2025 to 20% in 2026.

06

Outcomes: productivity, returns and jobs

Owners who use AI mostly report gains, and few report cutting staff. All of these figures are self-reported, so they show perceived benefit rather than measured return.

OutcomeFindingSource
Productivity71% of AI-using small firms report increased productivityFederal Reserve SBCS
Quality and sales39% report better quality; 31% report higher salesFederal Reserve SBCS
Overall impact93% report a positive business impactGoldman Sachs
Return on investment52% of AI users report a return; 24% have not yet seen oneBluevine
Augment or replace87% say AI augments rather than replaces employees; 66% of Census-surveyed users use it only to augment tasksGoldman Sachs; Census
Employment2% of firms report AI-related employment decreases; 8% of owners say they reduced roles, 6% say they hiredCensus; BizBuySell

The Federal Reserve also found most AI-using firms saw no change in labor costs. The Census working paper finds firms with broader AI integration have stronger commercial performance, but that is a correlation: better-run firms may simply be the ones that adopt more.

07

Barriers to adoption

The obstacle differs by stage: non-users see no fit, would-be users cannot find the right tool or the time, and current users worry about accuracy.

  • Non-users: "not applicable." Of the Fed survey's firms with no plans to use AI (33%), over half said AI is not applicable to their business. Nearly half of Third District firms had no plans, and the top reason there was also inapplicability, with many also citing a preference not to use it or accuracy concerns.
  • Planners: tools and time. Firms planning to start within a year named finding tools that meet their needs (54%) and the time to implement or train staff (37%) as top challenges.
  • Users: accuracy and fit. Current users cited accuracy (46%) and adapting tools to business needs (43%) most often (Federal Reserve SBCS).
  • Skills and support. In the Goldman Sachs survey, 73% said more training and resources would help, and owners cited technical expertise, choosing tools and data privacy as barriers to deeper use.
  • Trust. Bluevine found 82% of owners face barriers to going deeper and 78% do not fully trust AI for basic tasks, though that poll was short and lender-sponsored.

Cost was not among the top challenges in the Fed survey, although other surveys cite it. The Minneapolis Fed adds that among firms not planning to use AI, roughly 20% have safety concerns and a similar share lack knowledge of what AI can do.

08

Implications

These recommendations are my interpretation of the evidence above, not findings from the surveys.

For local business owners

  1. Start with one recurring writing or admin task (customer replies, quotes, listings, social posts) where a general assistant already works and results are easy to check.
  2. Treat accuracy as the main risk. Users in the Fed survey named it first, so keep a human review step for anything customer-facing or financial.
  3. Write a one-page rule on what customer data may be pasted into AI tools before adding more of them.
  4. Measure one number (hours saved, response time, leads handled) for 30 days before paying for anything.

For chambers, lenders and small business development centers

  1. Lead with training and worked examples by trade. Non-users mostly say AI does not apply to them, and 73% of Goldman Sachs respondents asked for training.
  2. Focus outreach on firms under 20 employees and on food service, construction and trades, where Census data show adoption is lowest and flat.
  3. Help owners choose tools. "Finding tools that meet our needs" was the top challenge for firms planning to start.

For researchers

Track local-service sectors directly. Current surveys group them with larger firms or report them only through broad NAICS sectors, so a true "local business" adoption rate is not yet published.

09

Method, caveats and sources

This report draws on government statistics, Federal Reserve research and industry surveys published through September 2026. Figures are as of October 2, 2026; BTOS updates every two weeks, so newer readings may differ.

Caveats

  • No survey measures "local business" directly. Small employer firms, firms with fewer than 20 employees, and local-service sectors are used as proxies.
  • Most outcome data are self-reported. Owners reporting "productivity gains" or a "positive impact" are not measured returns, and the Census link between AI breadth and performance is a correlation.
  • Several surveys come from interested parties. Lenders, software vendors, a business-for-sale marketplace and a bank's program alumni network have different sampling and incentives from the Census Bureau and the Federal Reserve.
  • BTOS changed its question in November 2025. Before that it asked about AI use in producing goods or services, so earlier figures are not directly comparable.
  • Some widely repeated figures were left out. Claims such as "89% adoption" or "3.7x average ROI" appear on aggregator blogs without a traceable primary source here.

Sources

See how it applies to you

The research says where. We'll show you where in your business.

The studies above are about businesses in general. One focused conversation tells you where AI would help you most.

Book your assessment