AI adoption in SMEs is stalling — not because the technology isn't ready, but because most businesses are asking the wrong first question.
The most common mistake in AI adoption at SME scale isn't choosing the wrong tool. It's asking the wrong question.
"What can AI do for us?" is the wrong question. The right question is: "Where is this business losing time, money, or quality through manual effort — and what's the commercial case for fixing it?"
That reframe changes everything. It anchors the conversation in business reality rather than technology possibility, and it forces every AI initiative to earn its commercial place before a line of code is written.
The commercial framework
I use a simple two-axis framework when evaluating AI use cases with a client.
Axis one: cost of the current state. What is the business actually spending — in time, money, or quality — on the problem this AI would address? Put a number on it. If you can't, the use case isn't ready to prioritise.
Axis two: ROI confidence. How confident are we that the AI intervention will produce a measurable improvement? High confidence means there's a proven approach, clean data, and a clear implementation path. Low confidence means we're in experimental territory.
Use cases in the top-right quadrant — high current cost, high ROI confidence — are where to start. They deliver fast, they're fundable, and they build the organisational confidence needed for the more ambitious bets later.
What the high-value use cases actually look like
In operationally complex SMEs, the highest-value AI use cases tend to cluster around a few specific patterns.
Document and data processing. Any process that involves extracting structured information from unstructured documents — job sheets, invoices, emails, application forms — is a strong AI candidate. The technology is mature, the data is usually available, and the manual cost is quantifiable.
Lead qualification and routing. Businesses that handle high volumes of inbound enquiries are spending skilled time on mechanical qualification. AI agents can handle the initial qualification conversation — capturing service type, budget, timeline, and location — and route to the right team with context attached.
Reporting and insight generation. The process of assembling, formatting, and distributing reports is a high-cost manual process in most SMEs. AI can automate narrative generation, anomaly flagging, and report distribution — freeing the people who understand the data to focus on acting on it.
The adoption reality
Technology adoption is never a technology problem. It's a behaviour change problem. The most sophisticated AI deployment in the world is worthless if the team uses it once and reverts to the spreadsheet.
Adoption requires three things: a clear process that the AI fits into, training that's specific to how the team works, and visible quick wins that build confidence. We build all three into every deployment.
AI isn't the answer to a question most SMEs haven't asked yet. The question is: where is the business losing commercial value to manual process? Answer that first.