Two well-known surveys, two different numbers
Two of the most frequently cited data points on small business AI adoption come from different organizations, different years, and different survey designs, and they don't measure quite the same thing. Goldman Sachs 10,000 Small Businesses Voices reported in March 2026 that 76% of small businesses in its survey currently use AI . The US Chamber of Commerce's Technology Engagement Center (C_TEC), in its fourth annual Empowering Small Business report released August 18, 2025, found that 58% of small businesses say they use generative AI, up from 40% in 2024 .
It's tempting to line these numbers up as a timeline — 58% in mid-2025, 76% less than a year later — and conclude that small business AI adoption jumped sharply in a few months. That inference doesn't hold up once you look at what each survey actually asked and whom it asked.
Further reference: Goldman Sachs press release, March 2026 survey; US Chamber: Latest Empowering Small Business Report press release
Different questions, different populations
The Chamber's 58% figure specifically measures generative AI use. Its original 2025 report says Teneo Research surveyed 3,870 US small businesses, defined as having fewer than 250 employees, between June 6 and June 26, 2025. Use that original methodology rather than treating a secondary summary as the source. Goldman Sachs' 2026 release describes a different sample and a broader AI-use label.
Goldman Sachs' respondents were 1,256 program participants surveyed from January 27 to February 4, 2026 by Babson College and David Binder Research. The sample is not the same as the Chamber's and the questions have different wording. We cannot determine the size or direction of the effect of those differences from the headline percentages alone.
Put together: one survey asks about AI in general and samples program participants; the other asks specifically about generative AI and samples a broader small business population. Neither question nor population matches, so neither does the resulting percentage.
Further reference: Goldman Sachs press release, March 2026 survey; U.S. Chamber: Empowering Small Business 2025 report (methodology included)
Why you can't build a trend line out of these two numbers
The Chamber itself reports generative AI use at 40% in 2024 and 58% in 2025. That is a comparison published within its own series. Readers should still inspect question wording and methodology before treating successive surveys as perfectly comparable. A press-release summary is not a complete account of every limitation.
Combining the Chamber's 58% with Goldman Sachs' 76% does not establish an 18-point increase. The honest conclusion is narrower: both report substantial use within their respective definitions and samples. They do not, together, measure a single national adoption trajectory.
The practical lesson for anyone citing these figures: always note which organization, which year, which exact question, and which population a small business AI statistic comes from, and resist stitching two differently defined figures into a single growth story. The Chamber report itself has three years of consistent methodology behind its trend claims; Goldman's report is a single snapshot that, so far, stands on its own.
- Goldman Sachs, March 2026: 76% report AI use; sample of 1,256 program participants.
- U.S. Chamber, 2025: 58% report generative AI use; 3,870 businesses with fewer than 250 employees.
- Different definitions, populations and fieldwork dates prevent a direct trend calculation.
Further reference: U.S. Chamber: Empowering Small Business 2025 report (methodology included)
Read five details before sharing a number
First, find the exact question: using any AI, using generative AI, trialling a tool and integrating it into operations are different measures. Second, identify who answered and who was eligible. Third, note fieldwork dates rather than only publication dates. Fourth, distinguish all respondents from the subset already using a tool. Fifth, check whether the result is behavior, opinion or an expectation.
A statement that users expect revenue growth is not measured revenue growth. A percentage among AI users is not a percentage among all businesses. When a chart loses those qualifiers, it can create a stronger story than the source supports. Retain the denominator and context even when the resulting sentence is less dramatic.
What this means for a small-business decision
Surveys can help identify recurring concerns, such as expertise, tool selection and data handling. They are less useful for selecting your first workflow. Start with your own work: how frequently does a task occur, how long does it take and how often does it require judgment? Those observations lead to a more defensible pilot than trying to match an adoption percentage.
Use a supplier checklist to turn broad concerns into questions. What sources will the tool use? Who checks inaccurate answers? What access is needed? How are updates handled? What is the full cost? You do not need to settle the national adoption rate to ask those practical questions.
Sources & further reading
- Goldman Sachs press release, March 2026 survey
- US Chamber: Latest Empowering Small Business Report press release
- U.S. Chamber: Empowering Small Business 2025 report (methodology included)
Hampala’s practical guidance, not legal advice. Examples are illustrative, not customer results. Research dates and survey populations are stated where relevant. Tools and provider terms may change.