The AI Tools Small Businesses Actually Pay For (And What They Skip)
Knowing that a small business "uses AI" doesn't say much. The more useful question is what it's actually being used for — and recent data on real spending and real business functions paints a much more specific picture than the adoption headlines do.
Where the Spending Actually Goes
The JPMorgan Chase Institute has the closest thing to ground truth here: rather than asking businesses to self-report, it tracks real Chase Business Banking transaction data for payments made to AI services. That data shows a small but meaningful split between "consistent" AI users — businesses paying for AI services steadily, month over month — and "sporadic" users who pay occasionally or inconsistently. The ratio of consistent to sporadic users has been climbing in recent years, suggesting that once a small business starts paying for AI regularly, it tends to stick with it rather than churn in and out.
The same data shows a generational pattern among adopters: businesses that started paying for AI more recently began at roughly half the initial monthly spending of the earliest adopters. That's consistent with cheaper, more accessible small business AI tools lowering the entry cost — a small business today doesn't need the same up-front investment an early adopter did in 2019 or 2020 to get started.
What They're Actually Using It For
The U.S. Census Bureau's Business Trends and Outlook Survey asked adopting firms which specific business functions they use AI in, out of 15 tracked categories — finance, HR, customer service, marketing, IT, R&D, and more. Two findings stand out. First, most businesses aren't spreading AI thin across their whole operation: 57% of adopting firms use it in three or fewer business functions. Second, when businesses do pick functions, they consistently gravitate toward the same few: Sales and Marketing (52% of adopters), Strategy and Business Development (45%), and IT (41%).
That's a meaningfully different picture than "AI is transforming every part of small business operations." In practice, it looks more like most small businesses that adopt AI pick one or two revenue-facing or planning functions and go deep there, rather than adopting it broadly across accounting, HR, legal, and operations all at once.
Where Adoption Is Highest — and What That Tells You
Federal Reserve analysis of the same Census data shows adoption is far from even across industries. Professional, scientific, and technical services and the financial sector lead, with adoption rates around 33% and 30% respectively — both knowledge-heavy sectors where the core "product" is often information, analysis, or advice, which maps naturally onto what generative AI tools are good at. Retail, hospitality, and construction — sectors more dependent on physical operations and in-person service — lag noticeably behind.
For a small business owner reading the adoption numbers and wondering why AI doesn't feel relevant yet, this is a useful sanity check: your industry's actual adoption curve may look nothing like the national average, and that's a data-driven reason rather than a sign you're behind.
The Practical Pattern
Put together, the picture that emerges from real spending and real function-level data is fairly consistent: small businesses that get value from AI tend to start with a specific, revenue-facing task — marketing content, sales support, planning — rather than a company-wide rollout, and they tend to keep paying for it once it's working rather than treating it as a one-time experiment. That's a more grounded starting point than "should I be using AI" — the better first question is which single function in your business looks like the ones already showing the highest adoption, and whether that's worth testing first.
Sources: JPMorgan Chase Institute, Small Business in the Age of AI series (2026); U.S. Census Bureau, Business Trends and Outlook Survey, AI supplement (Nov. 2025–Jan. 2026) and related Census working paper "The Microstructure of AI Diffusion"; Federal Reserve, "Monitoring AI Adoption in the US Economy" (April 2026).
*Disclaimer*: Harvard Business Services, Inc. is neither a law firm nor an accounting firm and, even in cases where the author is an attorney, or a tax professional, nothing in this article constitutes legal or tax advice. This article provides general commentary on, and analysis of, the subject addressed. We strongly advise that you consult an attorney or tax professional to receive legal or tax guidance tailored to your specific circumstances. Any action taken or not taken based on this article is at your own risk. If an article cites or provides a link to third-party sources or websites, Harvard Business Services, Inc. is not responsible for and makes no representations regarding such source’s content or accuracy. Opinions expressed in this article do not necessarily reflect those of Harvard Business Services, Inc.