Founder Data ReportA Four-Person Firm Almost Always Knows What It Is Doing Today. Six Months Out, One in Four Does Not.
You can’t afford to risk your business without a Sidekick at your side.
Census lets business owners answer “I don’t know” on exactly one subject. The smallest firms use it least of anyone about the present — and nearly three times as often when the same question is put six months ahead.
On the Census Bureau’s fortnightly survey of American businesses there is exactly one question an owner is allowed to answer “I don’t know” to: whether the business used artificial intelligence in the last two weeks. That makes it an accidental instrument. It does not measure what owners think about AI — it measures how far the answer sits from the person answering. In the fortnight to August 23, 2026, 8.5% of firms with one to four employees could not answer it. At firms with 250 or more employees, 24.5% could not.
- 8.5% of firms with 1-4 employees cannot say whether they used AI in the last two weeks; 24.5% of firms with 250+ cannot.
- Asked about six months from now, the smallest firms’ figure rises to 24.8% and the largest to 32.6%.
- The gap between largest and smallest falls from 16.0 points to 7.8 points once the question is about the future.
- The jump itself is almost identical at every size below 250 employees — about 16 points — and only half that at the largest firms.
- Being small buys you a clear view of the present. Most of that advantage is gone the moment the question turns to the future.
Nobody else could have answered it
A “don’t know” on this question is rarely ignorance about technology. It is distance. In a business of 250 people the person filling in the survey cannot see every department; somebody in marketing may well be running copy through a model without telling anyone. In a business of three, there is no department to hide in. The owner knows because the owner is the only one who could be doing it.
Bars sit on a common 0-35% axis so this chart and the next one can be read against each other. Orange is the one-to-four-employee class throughout both articles; slate is every larger class. The rise from 1-4 to 250 or more is 16.0 points and clears the survey’s own standard errors; the single step from 1-4 to 5-9 does not, so treat those two classes as the same.
Source: U.S. Census Bureau, Business Trends and Outlook Survey, cycle 202618, reference period August 10-23, 2026, published September 10, 2026; downloaded from census.gov/hfp/btos/downloads/ on September 19, 2026. Estimates are shares of responding employer firms, cut by employment size class.
Then the question moves to the future, and most of the advantage goes
The survey asks the same thing twice: whether the business used AI in the last fortnight, and whether it will be using AI six months from now. The first is a matter of record. The second is a forecast. For the smallest firms the share who cannot answer goes from 8.5% to 24.8%. For the largest it goes from 24.5% to 32.6%. Both rise — but the small firm’s answer rises from a floor and the large firm’s from a level it was already at, and the two end up much closer together than they started.
Each row is one employment size class on a common 0-35% track. The orange dot is the share who cannot say what the business is doing now, the slate dot the share who cannot say what it will be doing in six months, and the bar between them is what the question costs in certainty. Every one of the seven gaps clears the published standard errors.
Source: U.S. Census Bureau, Business Trends and Outlook Survey, cycle 202618, reference period August 10-23, 2026, published September 10, 2026; downloaded from census.gov/hfp/btos/downloads/ on September 19, 2026. Estimates are shares of responding employer firms, cut by employment size class.
You can see everything happening in your business today. That is not the same as being able to see six months.
The future costs almost everyone the same sixteen points
Measured as a gap rather than a ratio, the result is flatter than it looks. Moving the question from this fortnight to six months out adds about 16 percentage points of “don’t know” at every size class below 250 employees — 16.3 at one to four, 16.5 at twenty to forty-nine, 16.0 at fifty to ninety-nine. None of those differ from each other by more than the survey’s error allows. Only the largest firms break the pattern, at 8.1 points, and they do it because they had already spent most of that uncertainty on the present tense.
Percentage points, on a common 0-20 point axis. Each bar is that class’s six-month “don’t know” share minus its present-tense share. The six bars below 250 employees are statistically indistinguishable from one another; the 250-or-more bar is distinguishable from all of them. Standard errors here are combined as though the two questions were independent, which they are not — they are asked of the same firms in the same fortnight — so the real margins are narrower than the test allowed for.
Source: U.S. Census Bureau, Business Trends and Outlook Survey, cycle 202618, reference period August 10-23, 2026, published September 10, 2026; downloaded from census.gov/hfp/btos/downloads/ on September 19, 2026. Estimates are shares of responding employer firms, cut by employment size class.
What the number is actually telling you
The instinct is to read a low “don’t know” as competence and a high one as chaos. It is neither. It is a map of who has to hold the whole business in their head. The four-person firm answers because there is nobody else to ask, and that same fact is why the owner cannot delegate the answer, cannot check it against anyone, and cannot stop carrying it. The visibility and the burden are the same thing seen from two sides.
All three are from the same fortnight and the same question pair. The multiplier is the six-month share divided by the present-tense share within each size class, so it does not depend on how many firms are in the class.
Source: U.S. Census Bureau, Business Trends and Outlook Survey, cycle 202618, reference period August 10-23, 2026, published September 10, 2026; downloaded from census.gov/hfp/btos/downloads/ on September 19, 2026. Estimates are shares of responding employer firms, cut by employment size class.
Three limits worth stating. This is one fortnight, not a trend — though unlike the survey’s self-assessment questions, these two series do not step when Census redraws its sample each year, which was checked series by series before the comparison was made. The Census Bureau collected no data at all for the six weeks from October 6 to November 16, 2025 because of the funding lapse, so any longer series on this survey has a real hole in it rather than a smooth line. And the question is about artificial intelligence specifically: it is the only place the survey offers an “I don’t know”, which is what makes it useful here, but it is one subject and not a general measure of what owners know.
Frequently Asked Questions
How many small businesses do not know whether they use AI?
In the U.S. Census Bureau's Business Trends and Outlook Survey for the fortnight ending August 23, 2026, 8.5% of firms with one to four employees answered “do not know” when asked whether the business had used AI in the previous two weeks. Nationally the figure across all employer firms was 9.6%.
Do larger companies know more about their own operations?
On this measure, less. 24.5% of firms with 250 or more employees could not say whether the business used AI in the last two weeks, against 8.5% of firms with one to four. The likeliest reason is span rather than attention: one respondent at a large firm cannot see every function, while a four-person firm has no function the owner is not already in.
Are small business owners more or less certain about the future than large ones?
Less certain than they are about their own present, and still slightly more certain than large firms in absolute terms. Asked whether they will be using AI in six months, 24.8% of the smallest firms cannot say, against 32.6% of the largest. The gap between the two narrows from 16.0 points on the present-tense question to 7.8 points on the forward-looking one.
What is the Business Trends and Outlook Survey?
A U.S. Census Bureau survey of American employer businesses run every two weeks, covering current performance, revenue, employment, prices, supply chains and expectations six months ahead. Results are published by sector, by state and by employment size class. It has run since 2023; no data were collected for the reference periods between October 6 and November 16, 2025 because of the federal funding lapse.
Does not knowing whether you use AI mean a business is badly run?
It is not a competence measure. The share answering “do not know” rises steadily with headcount, which points at organisational distance between the person answering and the work rather than at how well the business is run. A small firm's low figure reflects that the owner is close enough to every part of the business to answer — which is also why nobody else can.
Published by The Lonely Entrepreneur — the community and coaching platform for entrepreneurs who are building alone. lonelyentrepreneur.com
This article is for educational purposes and is not a substitute for professional financial or legal advice.