Most AI projects don't fail. They quietly stop.

Most AI projects in small businesses do not fail. They just quietly stop.

Someone gets enthusiastic, a licence gets bought, two or three people try it for a fortnight, and then it fades. Nobody declares it a failure because nobody declared it a project. Six months later the subscription is still being paid and nobody can name a thing that changed.

The reason is almost never the technology. The tools are genuinely good now. What goes wrong is everything around them, and it is usually one of the same seven things.

It is not a single question

The temptation is to ask "are we ready for AI?" as though there were one answer. There is not, and treating it as one number hides the actual problem.

A business can have excellent tooling, because it already pays for Microsoft 365 Business Premium and has Copilot sitting there, while having no governance at all, so nobody knows what they may safely paste into it. Another business can have a tidy, well-organised document estate and no idea what its licences already include. Both would score the same on a single scale. They need completely different advice.

So it is worth separating them out.

Strategy and leadership. Is there one person accountable, are there two named business problems to attack, and is there any budget at all? "Explore AI" is not a goal. It is what people write down when they have not decided anything.

Data and documentation. Whether your information is findable and current, or spread across a file server, three OneDrives, an email archive and one person's desktop. This one matters more than anything else on the list and gets the least attention.

Tooling and licences. Whether you know what you already own. Most businesses on Business Premium are paying for capability they have never switched on, and buy something new instead.

Skills and confidence. Whether ordinary staff have had any practical training on their own work, rather than a demo. And whether they understand where these tools are unreliable, which is mainly arithmetic and citations.

Governance and policy. Whether there is anything in writing about what may and may not be put into an AI tool. Your staff are already using them. The only question is whether they know the rules.

Process readiness. Whether you can name your three most repetitive processes and roughly what they cost you in hours. Without a measured baseline you cannot prove a saving, so funding dries up.

Security and compliance. Whether permissions are in order before you point AI search at your files. The most common unpleasant surprise is not a wrong answer. It is an accurate one, from a document the person asking should never have been able to open.

The two that hold everyone back

If we had to bet on which dimensions are weakest in a business we have not met, it would be data and governance, every time.

Data, because it is unglamorous. Consolidating a document estate is weeks of tedious work with no demo at the end of it, so it never gets prioritised. But AI grounded on your own content is where most of the value actually is, and it is the one thing you cannot buy. Point a good tool at a mess and you get confident, well-written, wrong answers, which is worse than no answer at all.

Governance, because it feels premature. It is not. Staff are pasting things into free AI tools today, in every business, including yours. A one-page acceptable use policy that says plainly what must never go in is a day of work and removes a genuine data protection exposure. Twenty pages that nobody reads is worse than the one page.

Find out where you stand

We built an assessment that works through all seven, one question at a time.

Take the AI Readiness Assessment

Twenty-one questions, about four minutes. You answer on a four-point scale rather than yes or no, because maturity is rarely binary.

What comes back is a written report covering four things: your maturity level from 1 to 5, a score for each of the seven dimensions so you can see which are dragging, three priority projects drawn from your weakest areas, and a phased roadmap split into the next three months, three to six, and six to twelve.

There is also an ROI estimate, and it is deliberately scaled to your maturity level. A business at level 2 cannot capture in a year what a level 4 business can, and quoting both the same number would be flattering and useless. Work through the roadmap and the figure moves. That is rather the point.

The report arrives by email rather than appearing on screen, so use an address you can get to.

What to do with the result

Whatever level comes back, the advice is the same: do one thing.

The businesses that get value from AI pick a single process, measure how long it takes today, apply a tool to it, and measure again. Then they do the next one. The businesses that get nothing announce a transformation programme, buy several licences, form a working group, and have nothing running a year later.

If you already know roughly where you stand and just want ideas, our AI Use Case Finder gives you a ranked shortlist for your sector and size. If you want to know what it might be worth in pounds, the AI ROI Calculator does that. And if you would rather talk it through with someone who has done it in businesses like yours, get in touch and we will be honest about whether it is worth your time yet.