Guides

AI change management for a UK SME: getting the tools used, not just bought.

By Zain M · 5 October 2026 · 11 min read

AI adoption in a UK SME is decided by six things, none of them the software: a one-page policy with examples of what may and may not go in; shared workspaces or projects built on the firm’s own templates so the tool already knows how the firm works; one ninety-minute session per team on that team’s two most common tasks; a named owner who is visibly using it; a first measured win that is shared; and an honest answer on what happens to jobs. DSIT found 60 per cent of UK businesses cite limited AI skills as a barrier and the ONS found most firms with a skills gap are training their own staff rather than hiring. Firms that do the six things get most seats used within a month; firms that send a login get a third.

The barrierSkills, not software
The six leversPolicy, templates, session, owner, first win, honesty
TimeframeMost seats used within a month

What the UK evidence says about the barrier

DSIT’s AI Adoption Research found 60 per cent of UK businesses cite limited AI skills as a barrier to adoption, and the ONS found that 62 per cent of firms citing an expertise gap are training their own staff rather than hiring help. The July 2026 survey of recruitment agencies we cite elsewhere found the return on AI rose sharply with the amount of training an agency had done. Together those say the same thing: the tool is bought quickly and the skills to use it are built slowly, and the gap between the two is where seats go unused and staff drift to personal accounts.

The six levers

In the order they should happen.

01A one-page policy with examples. Not “use AI responsibly” but which tools on which plan, what may go in, what may not, three examples of each, and the rule that a named person checks anything that leaves the firm. Staff use tools they are sure they are allowed to use.
02Templates before training. Two or three shared projects or custom GPTs built on the firm’s own documents, tone and templates, so the first thing a new user sees is the tool already knowing how the firm writes a proposal or answers a client. An empty chat box is where adoption dies.
03One session per team, on that team’s work. Ninety minutes, hands-on, on the two tasks the team does most, using their own documents, ending with an artefact they keep. A mixed all-hands demo is too shallow for some and too technical for others.
04A visible owner. Someone senior who uses the tool in the open, shares what worked and what did not, and answers the question nobody wants to ask in a meeting. Adoption follows what leaders do, not what they say.
05A first measured win, shared. One process, measured before and after, with the number told to everyone. It converts sceptics faster than any presentation, and it is the evidence the next budget needs.
06An honest answer on jobs. Staff will not adopt a tool they believe is replacing them. Say what the tool takes (the retyping, the chasing, the first draft) and what it does not (the judgement, the relationship, the decision), and mean it.

The first month, week by week

Week one: policy adopted, business plans bought, personal accounts cancelled, templates built. Week two: sessions for the two teams with the most to gain. Week three: the first measured process goes live with a person still deciding; the owner shares what they are using it for. Week four: the remaining teams’ sessions, the first numbers shared, and a short survey of who is using what. A firm that does that sequence sees most seats in use by the end of the month and knows exactly which are not and why.

What to measure

Seats active in the last seven days, from the admin console, which every business plan provides. Tasks per team where the tool is used, from the session follow-up. The measured win on the first process, in hours and pounds. And the number of staff still on personal accounts, which should be zero and rarely is in month one. Report the four monthly for a quarter, then stop, because by then the tool is just how work is done.

The mistakes

Buying seats and announcing them. Running one demo for everyone. Letting the enthusiast run it, so it becomes their thing rather than the firm’s. Skipping the policy and discovering the data question from a client. Measuring nothing, so the renewal decision is a matter of opinion. And promising that nobody’s job will change, which staff do not believe and which is not true; what is true is that the boring half of most jobs changes first.

Common questions

Why do staff not use the AI tools we bought?

Usually because nobody told them what they are allowed to put in, the tool arrives as an empty box, and no one showed them on their own work. A policy with examples, shared templates and one session per team fixes most of it.

How long does AI adoption take in a small firm?

Most seats in use within a month if the six levers are pulled in order. Without them, a third of seats get used and the rest lapse.

What training do staff need for AI?

A ninety-minute hands-on session per team on that team’s two most common tasks, using their own documents, ending with an artefact they keep. Not a general demo.

Should we tell staff AI will not affect their jobs?

No, because it is not true and they will not believe it. Say what the tool takes and what it does not, and involve the people whose work changes in deciding how.

What should we measure?

Active seats, tasks per team where the tool is used, the measured win on the first process, and the number still on personal accounts.

Who should own AI adoption in an SME?

A senior person who uses it in the open, not IT and not the most enthusiastic junior. Adoption follows what leaders visibly do.

Want the first month run for you?

We adopt the policy, build the templates on your own documents, run the sessions per team and measure the first process, at a fixed price.

Start a conversation →