I have now had this conversation with enough Northampton business owners that I can almost script it before it starts. Someone calls me, slightly frustrated, and says something like: "We signed up for ChatGPT, we did a training day, and honestly... nobody's really using it." Sometimes they add, "I'm not even sure what we're supposed to be using it for."
This is not a technology problem. It is not a budget problem. It is not even a skills problem, not really. It is a sequencing problem - and it is almost entirely predictable.
"Most businesses don't have an AI problem. They have a strategy problem that AI has made visible."
The bridge-building trap
There is a story I keep coming back to. A group of military trainees are handed a pile of materials and given one instruction: "Build a bridge across the river. You have four hours." They immediately get to work - grabbing timber, organising into teams, moving fast. Within thirty seconds, the instructor stops them. "You've all failed."
They hadn't asked why. Had they taken a moment to ask what the actual objective was, they might have discovered they just needed to get a vehicle to the other side - and wandered downstream to find a shallow crossing where a Land Rover could simply drive across. No bridge required.
This is exactly what I see businesses doing with AI. "We need a chatbot." "We need to automate our emails." "We need an AI tool that does... something... with data." And they're off. Subscribing to seventeen different platforms. Asking ChatGPT to write LinkedIn posts whilst simultaneously wondering why the team isn't engaged.
The tools are not the problem. The sequence is.
The three mistakes I see most often
Mistake 01
Starting with tools instead of problems
The most common pattern: someone in the business reads an article about AI, buys a subscription, and then tries to find a use for it. This is backwards. The right question is not "what can AI do?" but "where in our operation are we losing the most time, making the most errors, or producing the most inconsistent output?" Start there. The tool selection follows naturally once you have a clear problem.
Mistake 02
Treating adoption as an afterthought
A training session is not adoption. I have seen businesses spend a full day with a trainer, come out with a list of prompts, and be back to their old workflows within a fortnight. Adoption is the work. Not the tools, not the prompts, not the subscriptions. Eighty percent of my job in the first three months of any engagement is helping teams actually use what is already built - building the habit, removing the friction, making it feel like their system rather than something imposed on them.
Mistake 03
Skipping the 10-80-10
I use a framework called the 10-80-10 rule. The first 10% is human: setting the context, the goals, the guardrails, understanding what you are actually trying to achieve. The middle 80% is AI - the heavy lifting, the repetitive work, the stuff that used to take hours. The final 10% is human oversight: the expert eye at the end checking whether the output is actually right and can be trusted. Most businesses that struggle with AI are skipping the first 10% entirely. They hand the task to the tool without giving it the context it needs to do good work, then wonder why the output is generic.
What actually works
The businesses I have seen make real progress in the first six months share a few things in common. They start with one workflow, not ten. They pick the thing that is eating the most time - usually something in sales, marketing, or operations admin - and they build a working system around that single problem before touching anything else.
They also involve the team early. Not as passive recipients of a new tool, but as active participants in designing how the tool fits their work. The people doing the job every day know where the friction is. When they help design the solution, they own it - and ownership is the only reliable predictor of consistent use.
And they measure it. Not in vague terms like "we're more efficient now," but in actual hours. How long did this task take before? How long does it take now? What is the weekly saving? When you can see the number, the habit sticks. When it is abstract, it fades.
The honest maths
Here is a rough calculation I run with most new clients. A typical 50-person Northampton business has somewhere between eight and fifteen people doing work that involves significant repetitive admin - writing, summarising, formatting, researching, drafting. If each of those people saves ninety minutes a day through well-built AI workflows, that is roughly 1,000 hours a month recovered across the business. At an average blended cost of £25 per hour, that is £25,000 a month in time value. Not cash in the bank, but time redirected to higher-value work.
The Foundation Programme costs £4,500 a month for three months. The maths tends to work out in the first month, if the implementation is done properly.
The caveat: none of this happens automatically. It requires the right sequence, the right workflows, and genuine commitment from the leadership team to see it through. The businesses that get the ROI are the ones that treat AI implementation as a project, not a subscription.
Where to start if you are in this position right now
If you recognise the pattern I described at the start - subscriptions bought, training done, nobody using it consistently - the first thing to do is stop adding more tools. Pause. Take stock of what you actually have and what problem you were originally trying to solve.
Then pick one workflow. The most painful one. The one that eats the most time, produces the most inconsistency, or creates the most frustration for your team. Build a working system around that one thing. Get the team using it every day. Measure the time saved. Then, and only then, move to the next one.
If you want a structured way to do this, the Discovery Workshop is designed precisely for this situation. Ninety minutes with your leadership team. We map your operation, identify the three highest-leverage AI opportunities, and you leave with a written Opportunity Map showing the recommended sequence. If you go on to the Foundation Programme within thirty days, the £750 is credited against your first month.
The shallow crossing is usually closer than you think. You just have to stop building the bridge long enough to look for it.
Damian
Founder, Rethinking Business · AI implementation for Northampton SMBs