Why AI training doesn't stick (and the "waves in the ocean" model that does)

Short answer: AI training fails in small businesses because it's treated as a one-off event, a single session, then nothing. Nick Harding, CEO of Fifty One Degrees, calls what actually works "waves in the ocean": training that keeps coming back, in different forms, again and again, rather than a single tick-box exercise.

Skip this, and the pattern is depressingly predictable. A few people use AI occasionally. Most go back to what they know. The business misses out on the productivity gains it was promised.

The mistake: treating training as a single event

Most small businesses that try AI training do the same thing. They book a session, get everyone in a room (or on a call), run through the basics of ChatGPT or Claude, and consider the job done.

It doesn't work. Nick Harding has seen this pattern repeatedly with UK small business clients, and he's blunt about why it fails: "This is not a once-and-done thing. This is something that isn't always-on activity that you basically need to think about again and again and again in various different mechanisms."

He compares it to waves in the ocean: "You just have to have wave after wave after wave of AI adoption."

That single sentence explains most failed AI rollouts in small businesses. The tools get bought. One training session happens. And then nothing reinforces it.

What happens if you don't keep the waves coming

Nick describes the failure mode plainly: "If you don't put that effort in, then like many other things, it will come along and fade away again, and you'll end up six months later asking yourself why you've been paying for licenses and nobody's using it."

That's a specific, avoidable outcome, not a vague warning. It's the gap between buying a tool and building a habit. Buying Claude or ChatGPT licenses for your team costs money every month whether people use them or not. If training doesn't stick, you're paying for software nobody opens.

What actually works: four concrete mechanisms

Nick doesn't just diagnose the problem. He sets out what to do instead, and it's specific enough to act on this month.

1. Run training in person, in the room

"I very strongly advocate you try to do in-person, in-the-room training with all of your team," Nick says. Video calls and pre-recorded modules are easier to organise, but they're also easier to half-watch. In-person training forces attention in a way remote sessions rarely do, and it gives people the chance to ask questions on the spot.

2. Build in a backup, something people can return to

Training happens once, but questions come up for weeks afterwards. Nick's approach: "We build the training modules into the AI platforms that we use. So if any of the team members have a question afterwards, they can go to that module in the AI platform and ask the questions there."

The practical version of this for most small businesses: keep a simple internal reference (a shared doc, a pinned Slack channel post, even a one-page PDF) that people can go back to without needing to interrupt a colleague or wait for the next session.

3. Rerun it periodically, don't assume it landed the first time

"You need to rerun that training periodically," Nick says. "Some people just won't have kept up. Other people may be off. Other people might be new to the business."

This is easy to skip because it feels repetitive to the person running it. But the audience for the second session isn't the same as the first: new starters, people who missed it, people who need a refresher. Build a rerun into your calendar now, three to six months out, rather than waiting until usage has already dropped off.

4. Weave AI into everything else you already do

This is the part most businesses miss entirely. Nick's advice: "Build AI messaging into all of those moments where you get together as a team. It might be an off-site, an away day, or you just go into the pub on a Wednesday evening. Any kind of other communications you send out, a quarterly update or some other sort of information, build it into those things. Build it into your performance framework if you have one and your training frameworks."

The point isn't to constantly lecture your team about AI. It's to make it a normal, recurring topic rather than something that only gets mentioned once, in one dedicated session, and never again.

Leadership has to model it too

Nick is explicit that this can't just be delegated to whoever runs the training session. "Make sure your leaders are also talking about this. This is something you really need to have permeated through everybody in the business. Leading from the top with whoever is implementing this, but also everybody else across the leadership team."

If your team never sees the owner or senior managers actually using AI tools themselves, no amount of formal training will convince them it matters.

Being honest about the effort involved

This isn't a quick fix, and it's worth saying so plainly: none of this works if you're not willing to put in ongoing effort. Nick's own framing makes that clear: "this requires always-on effort across various different disciplines." If you're hoping for a single afternoon that solves your team's AI skills gap permanently, this approach won't feel like a shortcut. It's the opposite of a shortcut. But it's also the difference between licenses that get used and licenses that quietly get cancelled a year later.

What to try this month

Pick one part of this you're not currently doing (a rerun date on the calendar, a shared reference doc, or a five-minute AI mention at your next team meeting) and start there. You don't need all four mechanisms running from day one. You need to stop treating training as a single event.

For a free, structured way to plan this out, the AI for Growth community gives you access to the Reskilling Navigator tool and a peer network of other small business owners working through the same challenge, without anyone trying to sell you anything.