The short version: A quarterly report published today from OnDeck and Ocrolus found that 61% of small businesses are now using AI, up from 58% last quarter, and 91% of those users say it has had a positive impact. Buried in the same report is a number almost nobody will quote: the median small business now takes in $1.0030 for every dollar it spends. Using AI in small business has gone mainstream at exactly the moment margins crossed back above break-even by three-tenths of one percent. Those two facts belong in the same sentence, and this piece is about what happens when you put them there.
Most coverage of this report will lead with the adoption number. That is the least interesting thing in it.
The Small Business Cash Flow Trend Report is not an opinion survey dressed up as research. It pairs responses from 805 small businesses holding working capital loans with anonymized cash flow data drawn from more than 3.76 million financing applications over 15 months. The survey ran June 16 to 24, 2026. That structure is unusual and worth pausing on: the same population that answers the questions also shows up in the transaction data. You get what owners say, and you get what their bank statements say, side by side.
So when 91% of AI users report a positive impact, there is, for once, a control group sitting right next to it.
What did the Q2 2026 report actually find?
The headline figures, per the report announcement:
- 61% of small businesses report using AI, up from 58% in Q1
- 91% of those users report a positive impact, up from 89% in Q1
- 93% of owners expect moderate to significant growth over the next year
- Inflation reclaimed the top spot as the leading concern at 34%, passing cash flow at 30%
- 75% of small businesses bypassed traditional banks for capital in Q2
- Median monthly non-bank loan inflows hit $8,885, up 11% year over year
And the one that anchors this article: the median revenue-to-expense ratio across all industries reached 100.30%, up from 99.84% in Q1.
Read that ratio slowly. In the first quarter of this year, the median small business in this dataset was spending slightly more than it earned. It was, on a monthly basis, running at a small loss. This quarter it crossed back over the line. The margin of victory is three-tenths of one percent.
Cory Kampfer, Co-President of Small Business Lending at Enova, framed the AI finding this way in the announcement: “It is no longer a novelty.” He is right, and the cash flow data is what makes that claim more than a vendor talking point.
Why doesn’t 91% satisfaction show up in the margin?
Here is the part worth thinking about. Line up the last three of these reports and watch both curves move together:
Q4 2025: 56% using AI, 87% reporting positive impact.
Q1 2026: 58% using AI, 89% reporting positive impact.
Q2 2026: 61% using AI, 91% reporting positive impact.
Adoption climbed five points across three quarters. Satisfaction climbed four. They moved in near lockstep.
That pattern is strange, and it is the most interesting thing in the report. When a technology crosses from early adopters into the mainstream, reported satisfaction usually dips. The first wave is self-selected: motivated, technical, patient with rough edges. The next wave is not. They arrive with less time, less appetite for tinkering, and higher expectations. Satisfaction normally sags as the tool meets people who did not go looking for it.
Here it went up instead. There are two honest readings, and they are not mutually exclusive.
The optimistic one: the tools genuinely got easier. Anyone who tried to get useful work out of a chatbot in 2023 and tried again this year knows the difference is not subtle. Later adopters may simply be having a better first experience than early adopters did.
The skeptical one: “positive impact” is measuring relief, not results. It is a feeling, and feelings are cheap to report. An owner who spent 40 minutes on a quote last year and 15 minutes on it this year will answer yes to that question, correctly and sincerely, whether or not those 25 minutes ever turned into anything that reached a bank statement.
The margin data does not settle the argument, but it does bound it. If AI were producing large, broad, bottom-line gains across 61% of small businesses, a median revenue-to-expense ratio of 100.30% is not what that would look like. What it looks like instead is a genuine improvement, arriving into a cost environment that is eating most of it. Inflation did not overtake cash flow as the top concern by accident.
That is not a case against using AI in small business. It is a case for being specific about what you expect from it.
Which AI tools are small businesses actually using?
The tool-share data in this report shifted more in one quarter than the adoption number did. ChatGPT remains dominant at 81% of AI users, but that is down from 90% in Q1. Claude more than doubled quarter over quarter to 33%.
Two things follow from that, and neither is “switch tools.”
First, owners are running more than one. Those percentages sum well past 100, which means a meaningful share of small businesses now keep two or more assistants open and route work between them. That is a maturity signal. It is what people do once they have learned that different tools are better at different jobs.
Second, the incumbent’s share fell nine points in three months without the incumbent doing anything wrong. That is what a competitive market looks like, and it is why we have argued that this is a poor season to sign a long annual contract. When share moves that fast, pricing and capability move with it.
What should you do with this?
If you are in the 61%, the useful exercise this quarter is to convert one instance of “positive impact” into one number that appears on a bank statement.
Not a dashboard. Not a time-savings estimate built from a guess about how long something used to take. One number, of the kind your accountant would recognize. Candidates that actually work for most small businesses:
- Quote turnaround time. If AI drafting cut your average quote from two days to four hours, your close rate on those quotes is the number. It is already in your CRM or your invoicing tool.
- Invoices sent per week. Slow invoicing is a cash flow problem disguised as an admin problem. If automation moved this, it shows up in your receivables aging within a month.
- Jobs completed without adding hours. This is the empowerment number, and it is the honest version of the AI productivity claim. Same team, same payroll, more work out the door.
That last one deserves emphasis, because the surrounding conversation about AI keeps sliding toward headcount. The data here does not support that framing and neither do we. In a quarter where the median business is clearing three-tenths of a cent per dollar, the win is not a smaller team. It is the same team absorbing more volume without burning out, and capacity you did not have to hire for. That is a margin story and a workload story at the same time.
If you want a structured way to run this, we walked through a lightweight version of OpenAI’s four-part ROI scorecard that any owner can complete without a finance team. And if you have not yet decided where to point AI first, the sequencing question matters more than the tool choice.
One caution worth carrying out of this report. Adoption at 61% with satisfaction at 91% describes a lot of businesses using AI enthusiastically and informally. We have written before about how far ahead of written policy that adoption has run, and nothing in this quarter’s data suggests the gap has closed.
Frequently Asked Questions
What percentage of small businesses use AI in 2026?
According to the Q2 2026 Small Business Cash Flow Trend Report from OnDeck and Ocrolus, 61% of small businesses report using AI, up from 58% in Q1 2026 and 56% in the quarter before that. The figure comes from a survey of 805 small businesses holding working capital loans, conducted June 16 to 24, 2026.
Does using AI in small business actually improve profitability?
The evidence so far is encouraging but modest. In the same report, 91% of AI users said the technology had a positive impact, yet the median revenue-to-expense ratio across all industries sat at 100.30%, meaning the typical small business earns just over a dollar for every dollar it spends. AI appears to be helping, but rising costs are absorbing much of the gain, so owners should tie AI use to a specific measurable outcome rather than assume it reaches the bottom line on its own.
What is a revenue-to-expense ratio and why does it matter?
It is monthly revenue divided by monthly expenses, expressed as a percentage. Above 100% means a business is taking in more than it spends; below 100% means it is running at a loss. The Q2 2026 figure of 100.30% is notable because it marks a return above break-even after Q1 came in at 99.84%, and because a margin that thin leaves very little room to absorb a bad month.
Which AI tools are small businesses using most?
ChatGPT remains the most widely used at 81% of AI users, though that is down from 90% in the prior quarter. Claude more than doubled its share quarter over quarter to reach 33%. Because those figures total well over 100%, many small businesses are clearly using more than one assistant and routing different kinds of work to each.
We keep coming back to the same question, and this report sharpened it: if you are using AI in your business right now, can you name the one number on your books that it moved? We would genuinely like to know which number you picked, and whether it held up over a full quarter.
