How Much Time AI Saves at Work, According to the Census Bureau

Something has been working. Over the last two years the tools have taken real weight off small businesses: the first draft that used to stall a Tuesday, the research pass, the pile of receipts, the summary of a meeting nobody had time to write up. If you have felt that, you were feeling something measurable.

You may also have quietly wondered whether you were getting the small version of it, because the story around you was much larger. In March 2026 the US Census Bureau put a number on the ordinary version.

The Census Bureau's Household Trends and Outlook Pulse Survey found 56% of US workers had used AI on the job for at least one of eleven measured tasks. Among those who used it in the previous week, a quarter saved less than an hour and 31% saved one to two hours. Counting the 10% who said it saved them no time, about two-thirds of that week's users recovered two hours or less.

What workers used AI for

Four of the five sit on the shallow side of the work. On generating ideas, a tool can draft the options and you still choose among them, which is where your taste comes in.

Why one to two hours reads as a disappointment

The number lands badly because of what it's being compared to.

Is the story you are measuring against a measurement, or a projection? Is it describing a firm of your size, with your data, doing your work? Is the person telling it selling something whose value depends on the number being large?

If you got an hour or two back, you're in the same band as nearly a third of the people the Census asked who had used AI that week.

In our reading, an hour or two is about what one handed-over task gives back, and the gain adds up once you hand over a second.

What seventy hours looks like

At an hour and a half a week, you're looking at roughly seventy hours a year per person, the better part of two working weeks.

Seventy hours is a real return and a modest one; it's too small for a hire. You could put it toward the concept work you keep pushing to the evening, or the follow-up calls that decide next quarter, if you decide in advance where the hours go. Without that decision, returned time tends to evaporate, which is the subject of the productivity paradox briefing.

You get the compounding from repetition. Hand over one recurring task this quarter, watch it for a month, then hand over the next one. Four quarters of that is four tasks off the desk, and each one was small enough that nobody had to be convinced.

From time spent on due diligence for new clients or turning over your notes from initial engagements into a full team brief, our firm has saved 4-8 hours of time per week for creative businesses to streamline their workflows and focus on the more important work that contributes to the bottom line.

What you can do this week

Count your own, for one week, and count it honestly.

Keep a list open and log the minutes at the moment you finish a task, with the task named. Searching, drafting, summarizing, admin, whatever it was. By the end of the week you'll have your own version of the Census table, drawn from your own business.

The failure in week two is that the small entries never get logged. Twenty minutes here and ten there feel too minor to write down, and they're the entries most likely to go missing from the total. Log each one when you finish it; a list written up at the end of the day comes out low, and you'll draw the wrong conclusion from your own data.

Then pick the single task that appeared most often, and make that one repeatable before you touch anything else.

All Why Adoption Stalls Briefings