Eighty-six percent of small-business owners in the US now use AI, up from 56 percent just two years ago. Only 15 percent say it has actually changed how their business runs.

That gap is the real story. A survey of graduates of Goldman Sachs' 10,000 Small Businesses programme, run with Babson College (2,473 responses, fielded 10 to 24 August 2026), found adoption growing faster than almost any business tool in memory. The US Federal Reserve's own 2026 Small Business Credit Survey, a much larger sample of 6,525 US employer firms fielded through late 2025, tells the quieter half of the same story: 46 percent of owners now use AI, but among those users only 7 percent say it is built into how the business actually runs. Most are still experimenting.

If you've signed up for a chatbot, used it to draft a few emails and tidy some meeting notes, and felt quietly modern about it, you are not behind. You are exactly on trend. You are also, almost certainly, one of the majority who haven't changed anything that actually moves the numbers yet.

Call it adoption theatre. It is the single most common mistake I see SME owners make with AI right now, and it is not refusing to use it. It is using it in a way that changes nothing underneath.

The pattern is always the same. An owner, or their best staff member, starts using AI to speed up a task they already do: drafting an email, summarising a document, tidying meeting notes. It feels faster. It is faster. But the workflow around that one task, the handoffs, the approvals, the places where work sits waiting on a human, hasn't moved. The business still runs at the same speed it always did, because the constraint was never the typing. It was everything around the typing.

This isn't a technology problem. It's the natural result of bolting a new tool onto an old process instead of asking what the process should look like if you built it around the tool from scratch. Almost every owner I coach through this makes the same assumption going in: that using AI more often, on more tasks, is the goal. It isn't. Redesigning one workflow properly is worth more than using AI lightly across twenty of them.

Here's the fix, and it's smaller than most owners expect.

1. Pick one repeatable workflow. Not "try AI on everything." One thing your business does every week, quoting, onboarding, follow-up, reporting, scheduling. Pick the one that currently eats the most hours or causes the most friction.

2. Redesign it around AI, from the first step. Don't ask "where can I add AI to this?" Ask "if I were building this workflow today, with AI available from day one, what would it actually look like?" Usually, two or three of the old steps simply disappear.

3. Measure the hours it frees, and reinvest them in the one thing only you can do. Before touching a second workflow, put the time you just freed into the next sale, the next hire, or the decision that's been sitting in your inbox for a month. That's where the return shows up on the profit and loss, not in the subscription you just signed up for.

"What we achieved in the last 3 months working with Kevin would have taken 2 years. Kevin's assistance has been very inspirational to our staff."

— Chris, Global Culture

Redesigning a workflow takes something most owners are already out of: uninterrupted time, and the nerve to change something that's currently, if slowly, working. It's far easier to layer a tool on top of the existing process than to stop, map the workflow properly, and rebuild it. Nobody else in the business is going to push for that redesign either, staff are usually relieved just to have an easier version of the task they already do.

That's the real execution gap behind the 15-versus-85 split. It isn't a lack of AI skill. It's the lack of a structured, outside-led process for deciding which one workflow to redesign first, and the accountability to actually finish it in a quarter instead of drifting to the next shiny feature. This is usually exactly the gap coaching closes: not teaching the tool, but making sure the one redesign that matters gets done properly.

This is also where Simple-Growth's expertise ends and a specialist begins: which specific AI tool to trust with customer data, and how to vet it for privacy and security, is a decision for a qualified IT provider, not a business coach. CERT NZ is a sensible starting point for NZ owners weighing that call, and knowing that boundary is part of building a properly supported business.

The one workflow you redesign properly this quarter will do more for your business than the next ten AI subscriptions combined.