The single most confusing data point in small business AI coverage is that two credible surveys, asking adjacent questions, have produced wildly different numbers. The SBA Office of Advocacy research spotlight put small business AI use at roughly 8.8% by August 2025. The SBE Council 2025 survey reported that 88% of small businesses use AI tools. Both can be defended. They are answering different questions.
The SBA figure measures whether AI is integrated into business operations in a way the firm itself recognises as meaningful. The SBE Council figure includes any owner who has typed a question into ChatGPT in the last quarter. The accurate picture sits somewhere in between, and the more interesting trend is that the gap with large firms is narrowing in both data sets. Thryv’s 2025 small business survey found AI adoption rising from 39% to 55% year over year, roughly a 41% relative increase, and Goldman Sachs’ 10,000 Small Businesses survey put owner usage at 68% with another 9% planning to start within a year.
Whatever the exact number, the trajectory is the same. Small business owners are catching up to large enterprises faster than they have for any previous productivity technology, and the ones not yet catching up are increasingly visible by their absence rather than as the silent majority.
What owners are actually using AI for
The use cases sort into four categories.
Marketing is the largest. Drafting social posts, generating image variants for ads, writing email newsletters, refreshing website copy, and producing content faster than a one-or-two-person business has ever been able to. The owner who hated writing the monthly newsletter now generates a draft in ChatGPT or Claude and edits in 20 minutes. That single change is responsible for a meaningful share of reported AI satisfaction in the small business surveys.
Customer service is the second. Live-chat AI on the website, automated email response drafts, FAQ generation, and after-hours coverage that used to require a paid service or a missed conversation are now handled in-house. The cross-sector data on AI customer service shows roughly 65% of incoming queries resolved without human intervention in 2025, up from about 52% in 2023. Small businesses see a milder version of that effect because their volume is lower, but the principle holds.
Bookkeeping and financial admin is the third. AI-assisted categorisation in the bookkeeping software, AI-drafted client invoices, AI-summarised expense reports. The work that used to be the dreaded weekend catch-up for owners is now half-done by the time they sit down to it.
Content for sales and customer education is the fourth. Most small businesses have always known they should produce blog posts, case studies, and how-to videos. Most have not done it because the time cost was prohibitive. AI lowers the threshold enough that owners are doing it for the first time at scale.
The confidence gap is the most telling number
The most useful data point in the small business AI literature is not the adoption headline. It is the confidence gap. Across multiple 2024 to 2025 surveys, only around 27% of small businesses reported feeling confident adopting AI, compared to 82% of midsize firms. The adoption gap is closing. The confidence gap is wider, and it is the leading indicator of what happens next.
What that means in practice is that small business owners are using AI in a tentative, individual, prompt-by-prompt way, while midsize firms are running structured workflows. The owner who has tried ChatGPT on three things and found it useful is still nervous about putting customer data into it. The owner who has not tried it is afraid of looking foolish. Neither group is operating with the discipline that turns AI into compounding gains.
What changes when an owner gets the workflow right
The transition that matters is not the first prompt. It is the moment an owner stops asking AI for one thing at a time and starts treating it as a delegate they brief once and check on later. That looks like a written brand-voice document the owner hands the model on every task. It looks like a saved set of prompts for the recurring work (monthly newsletter, weekly social, quarterly review email, end-of-month bookkeeping summary). It looks like the owner spending less time generating content and more time editing it.
Owners who have made that transition consistently report meaningful weekly time savings. The “saves 10 to 12 hours a week” figures floating around small business AI marketing are not from controlled studies and the actual number depends heavily on the business mix. The closest peer-reviewed evidence on AI productivity for adjacent work is the Harvard Business School and BCG consulting study at 25% time reduction within scope and the GitHub Copilot RCT at 55.8% speedup on a benchmark task. The realistic gain on AI-good tasks for small business work sits in the 20 to 40% range, which on the four use cases above can recover several hours a week.
The SBA gap-closing finding is the headline
The SBA research spotlight’s most important sentence is not the 8.8% adoption number. It is the finding that the gap between small and large firm AI use is narrowing. Historically, small businesses adopted productivity technology three to five years after large enterprises did. With AI, the gap is closer to one to two years and shrinking. The cost is low (a $20 ChatGPT subscription is within almost any business’s marketing budget), the learning curve is forgiving, and the use cases are immediate.
That has not been true of any previous tech wave. Cloud accounting took close to a decade to penetrate small business at enterprise rates. CRM software was longer. Even basic websites took years. AI is moving differently because the ROI shows up in the first session, not the first year.
What the next 12 months look like
Three movements are likely to be visible by the end of 2026. First, the SBA-style “meaningful integration” measure will keep climbing past the current 8.8%. The Thryv-style “any usage” measure has already gotten close to ceiling and will plateau. The interesting metric to watch is the difference between the two, which is the share of small businesses that have moved from experimentation to operational use.
Second, the confidence gap will start to close, but unevenly. The owners who join peer groups, accountants who advise their small business clients on AI workflows, and software vendors who build AI into the tools owners already use will pull a meaningful share of the long tail forward. The owners who try to learn alone will continue to lag.
Third, AI-native small business tools will start displacing legacy SaaS in some categories. Bookkeeping, scheduling, and customer service software are particular pressure points. The vendors that build AI into the core flow rather than adding it as a feature flag will pull share.
If you own a small business, the question is not whether to start using AI. The question is whether you are going to let your direct competitors close the confidence gap before you do, or whether you are going to be the one a customer or a peer points at as the example of what running a business well looks like in 2026.