The headline number this year is genuinely big: roughly 68% of small businesses now use AI in some form, and among employers it's higher — 82% say they've invested in AI tools. Access is basically solved. Almost everyone has it open in a tab.
So here's the more useful question. If two-thirds of businesses use the same tools, why are a small slice clearly getting far more out of them? It isn't the subscription. It's how they use it.
The survey data spells out what "most" looks like: the typical small business uses ChatGPT or a similar tool for ad hoc tasks — draft an email, brainstorm some copy, summarize a document. One thing, one time, then close the tab. Useful, sure. But it's the AI equivalent of doing your books by retyping every number from scratch each month. You get an answer. You also get nothing you can reuse.
Ad hoc is the floor, not the finish line
Think about your own week. There's a set of things you do over and over: the quote follow-up, the weekly social posts, the "can you send me more info" reply, the recap after a call. If you're like most owners, you currently hand each of those to AI from a cold start — open a blank chat, re-explain your business, paste the context, fix the tone, ship it. Then tomorrow you do the whole thing again from zero.
The owners pulling ahead did one thing differently. The second time they noticed themselves typing the same kind of request, they stopped and built it once — a saved prompt, a template, a small assistant that already knows the business and produces the thing in their voice. The task went from fifteen minutes of wrangling to thirty seconds of review.
Stop asking AI to do a task. Start building the thing that does the task for you — every time, the same way.
Why this is where the money actually is
The returns show up once you cross from ad hoc to repeatable. Marketers who use AI report recovering around six hours a week on average; on content work specifically, small businesses report saving anywhere from five to fifteen hours. At a conservative $25 an hour, that's roughly $6,500 to $19,500 of time a year handed back to you. And it compounds — 91% of small businesses using AI say it boosts revenue, with most planning to spend more, not less.
But notice where those numbers come from. You don't recover six hours a week by asking a clever question once. You recover it by taking the five tasks you repeat every week and making each one a one-click job instead of a from-scratch job. The savings live in the repetition, which is exactly the part "winging it" leaves on the table.
The move this week
You don't need a new tool. You almost certainly already have one open. Do this instead: for the next few days, notice every time you ask AI for something you've essentially asked before. Write those down. By Friday you'll have a short list — probably three to five things — and that list is your build queue.
Pick the one you do most and turn it into something permanent: a saved prompt with your business baked in, a reusable template, a small assistant set up once. Then use it for a week and feel the difference between starting from a blank box and starting from something that already knows the job. That single shift — from a tool you reach for to a system you built — is the whole gap between the 68% and the few who are quietly running circles around them.