Within one week in early October, three credible organisations published AI adoption numbers for small business: 43%, 66% and 99%. Same country, same rough question, and a spread wide enough to cover almost any claim anyone wants to make.
If you run a small business and you are trying to work out whether you are behind, that spread is useless on its face. It becomes useful once you look at who was asked and what counted as using AI, because then the disagreement turns out to be about the top of the funnel rather than the bottom.
Every one of those figures was read from the publisher's own page rather than from coverage of it, and all four pages state their methods except the Chamber, which publishes the number without the sample size or the fieldwork dates.
The population is different every time. Heartland covers businesses up to 499 employees. TD covers owners up to 100 employees who also clear a revenue floor, which excludes the smallest and newest businesses, the ones least likely to have adopted anything. A survey of established, revenue-positive owners will always read higher than a survey of the whole small business population.
Using AI is not one thing. TD's sentence counts anyone "using it to save time, increase sales and improve cash flow", which can be true of a single person pasting a prompt into a free chatbot. SAS and IDC asked a different question entirely, about whether AI is embedded into strategy, operations and decision-making, and only 9% cleared that bar. Both statements can be true at once. One business that drafts its emails with a chatbot and one that runs its quoting process on an agent both count as adopters in the first measure and only the second counts in the second.
Sample size and recruitment decide how far a number travels. 200 respondents is enough to describe those 200 people and not enough to describe a country. The TD survey also recruited from an online panel of owners who chose to answer, which is normal commercial practice and also why its 99% should be read as "almost every confident owner we reached said yes" rather than "almost every small business uses AI".
The people answering differ. An owner answering a question about their own habits, a leader answering about their organisation, and a technology buyer answering about tooling are three different vantage points. The earlier look at the adoption gap covers the 66% against 9% split in detail; the point here is that the split is not a contradiction. It is a funnel.
Strip the top-of-funnel question away and the surveys converge far more than the headlines suggest. Use of some AI tool: somewhere between a large minority and most small businesses, depending on whom you ask. AI embedded into how the business runs: small, single digits in the SAS and IDC research, and consistent with the Chamber's own framing that using AI is not the same as being enabled by it.
That convergence is the part worth acting on, and it lines up with what shows up in why small business agent adoption lags enterprise and in businesses piloting AI before proving return. The scarcity is not access to tools. It is the operating discipline that turns a tool into a process, which is also why deployment rather than budget is the real barrier.
Survey percentages describe a population. These describe yours, and four of the five can be extracted without buying anything.
If you cannot fill in the first one, the survey you should trust is the one that says you are still experimenting, whatever the headline percentage.
Four questions, asked in this order, settle most of them. Who was asked, and does that match your business size and type. What counted as using AI, and would a single chatbot session qualify. How many people answered, and how were they recruited. Who paid for the research, which does not invalidate it and does change which findings get the headline.
Any report that answers all four is worth reading. A number without them is a headline rather than a measurement, and it is the reason AI for operational efficiency starts from a baseline in your own business rather than from an industry figure.
No. That figure describes 200 surveyed owners of established businesses who chose to answer an online panel. It is evidence about confident adopters, not about the whole population.
None of them as a statement about your business. Use them to understand the range and the direction, then measure your own functions. The published research is most useful for what it says about the shape of the problem, which is that use is common and embedding is rare.
Partly because adoption grows and partly because the bar for counting as adoption stays low while the tools get easier to reach. Both effects push the number up, and only the first one changes how a business runs.
It changes the target. If most businesses already use AI in some form, then using it is not an advantage. Embedding it into a function that runs without you is, and that is a narrower and more specific piece of work.
Maxpertise builds and runs the AI team that operates your business, in a managed lane where we host and run the team, or an embedded lane where AI engineers work inside your existing team. Both lanes keep the two human gates: a person approves the plan before the work starts, and a person reviews the result before it ships.
That structure exists because the interesting number is not how many businesses touch AI. It is how many functions run without the owner, and moving that number requires someone accountable for the process rather than a subscription. The work starts from a written scope you can read, close to the approach in spec-driven development with AI agents. If you want the method, start at how it works; the engagement options are on pricing, and contact with the two functions you would most like to stop running yourself is enough to scope it.
[1] Heartland Forward, Heartland small businesses match the nation on AI adoption as owners say AI boosts their bottom line, October 1, 2026. Read directly: "The survey of 691 small business owners nationwide, conducted by Echelon Insights and including an oversample of 308 small business owners in heartland states, found that 43% of small businesses currently use AI and that 70% of those businesses say it has had a net-positive effect on their finances."
[2] US Chamber of Commerce, 2026 Empowering Small Business report, October 2, 2026. Read directly: "2/3 of small businesses now use AI. (66%), up 8 points from 2025 and nearly 3 times the share using it in 2023 (23%)." The page does not state the sample size or the fieldwork dates.
[3] SAS, Majority of SMBs remain in early stages of AI maturity, May 14, 2026. Read directly: "Nearly 70% of small and midsized businesses (SMBs) still remain in the experimental or opportunistic stages of AI maturity", based on a survey of more than 1,600 SMB leaders across 28 countries, commissioned with IDC.
[4] SAS, AI readiness report for small and midsize businesses, accessed October 11, 2026. Read directly: "37% are still experimenting", "9% are fully embedded: Only 9% have fully embedded AI into strategy, operations and decision-making."
[5] TD Bank, TD survey finds small businesses confident despite economic headwinds and delayed investments, October 5, 2026. Read directly: "AI is becoming a practical business tool, with 99% using it to save time, increase sales and improve cash flow." The survey was conducted by Wakefield Research among 200 owners of businesses with 100 employees or fewer and revenue of at least 100,000 dollars, by online interview from August 17 to 28, 2026.
Maxpertise is an AI-native engineering company. We embed native AI engineers inside your team, live in about 10 days. Please enable JavaScript to view the site, or email contact@maxpertise.net.