The UK AI Productivity Gap: Why Small Businesses Are Using AI But Not Seeing Results

UK businesses have taken up AI faster than almost anyone expected. The government’s second AI adoption survey, published on 29 October 2026, found that 32% of businesses with five or more staff now use at least one AI technology, double the 16% recorded a year earlier.

The productivity numbers went the other way. Among those adopters, the share reporting higher employee productivity fell from 56% to 50%, and 42% said AI had made no difference to productivity at all.

That gap between adoption and results is now the central question for UK small businesses, and for a Treasury that is counting on AI to lift growth. Here is what the latest data shows, why so many firms are stuck, and what the businesses getting real returns are doing differently.

What The New Government Survey Found

The survey was commissioned by the former Department for Science, Innovation and Technology and published by the Department for Business, Innovation, Science and Trade. BMG Research interviewed 3,503 private sector businesses between January and April 2026, and the full report compares the results with the first wave carried out in 2025.

Measure	2025	2026
Businesses using at least one AI technology	16%	32%
Businesses that say AI is not relevant to them	51%	38%
Average share of staff using AI (adopters)	30%	31%
Adopters reporting higher employee productivity	56%	50%
Adopters reporting lower employee productivity	1%	4%

Adoption still rises with size. Some 27% of micro businesses (5 to 9 employees) use AI, against 34% of small businesses, 47% of medium-sized firms and 52% of large ones. Use is concentrated in simple tools: text generation and large language model tools are used by 30% of all businesses, mostly for administration, marketing and IT.

Spending is low, too. Of the businesses that could estimate it, 47% spent nothing at all on AI in 2025, relying on free tools or AI features built into software they already pay for. Only 14% of adopters rate AI as highly important to their operations, and 64% have seen no change in operating costs.

Other surveys tell the same story with different headline numbers. The British Chambers of Commerce put adoption at 54% of firms in March, and a Simply Business survey in October found 47% of small business owners now use AI, up from 22%, but only 19% feel very confident using it day to day. The methods differ, but the direction is consistent: more firms are trying AI, and far fewer are getting much out of it.

Why More AI Has Not Meant More Productivity

The research points to four reasons the gains are not showing up.

AI is being used at the edges of the business

The government report’s own summary is that AI is augmenting existing activities rather than changing how businesses operate. Most adopters use it to draft emails, summarise documents and tidy up marketing copy. Those are real time savings for the person doing them, but saving ten minutes on an email rarely changes how much a business produces. And the average share of staff using AI inside adopting firms barely moved, from 30% to 31%.

Few firms have redesigned the work around it

A report published on the same day by The Productivity Institute, The AI Ambition Gap, found that 89% of organisations use AI somewhere, yet only a quarter have redesigned work around it and 46% have moved beyond pilots. Just 6% earn at least 5% of their profits from AI.

The report’s most telling example comes from software teams. Programmers with AI assistants tripled their code output but released only 30% more software, because human checking and testing became the new bottleneck. Speeding up one step simply moves the queue to the next one. The same thing happens in a small business when AI drafts invoices faster but someone still has to rekey them into the accounting system.

Cleaning up AI output eats the savings

Zapier’s AI workslop survey of more than 1,100 US enterprise AI users found that the average employee spends 4.5 hours a week revising, correcting or redoing AI-generated work, even though 92% say AI makes them more productive. Workers who had been trained were six times more likely to report productivity gains. The survey covered large US firms, but the dynamic is familiar in small teams, where the owner is often the one checking every AI draft before it goes out.

Skills, not software, are the real constraint

Limited AI skills or knowledge affected 67% of UK businesses that had adopted AI and 78% of those planning to. Among non-adopters, 65% said they could not see a clear business case. When businesses talked about cost, they mostly meant the time and effort of implementing AI properly, not the price of the tools.

None of this is unique to Britain. A study of nearly 6,000 executives across the US, UK, Germany and Australia, published through the US National Bureau of Economic Research, found that around nine in ten firms reported no impact from AI on their productivity over the previous three years.

What The Businesses Getting Results Do Differently

The government survey picked out a small group it calls intensive adopters: firms that have embedded AI into their applications and processes or built their own tools, rate it as highly important and use it constantly. They make up just 2% of businesses surveyed, but their results look very different.

• 81% report higher employee productivity, against 50% of all adopters.
• 54% report lower operating costs, against 24% of all adopters.
• 86% use AI to automate tasks, against 58% of all adopters.
• AI reaches 60% of their staff on average, against 31%.
• They spent an average of £31,300 on AI in 2025, against £7,800 for adopters as a whole.

The group partly selects itself, since rating AI as important is one of the criteria, so these figures are not proof that spending more pays off. But the pattern matches other research. The Productivity Institute found that UK firms rated as well managed in 2020 were more likely to have adopted AI three years later, whatever their sector, size or location. The British Chambers of Commerce found that firms already using AI were far more optimistic about productivity than those still deciding, with a net balance of +71 points among current users.

The common thread is integration. The firms seeing results have connected AI to the systems where work actually happens, such as the CRM, the accounting package and the project management tool, rather than leaving it in a chat window on the side. That is also where specialist automation agencies and consultants tend to start: mapping how data moves between apps before adding AI to any step.

What The Autumn Budget Means For Small Firms

The 28 October Budget put AI at the centre of the government’s growth story, with a package worth more than £700 million. Most of it goes to AI security and building domestic AI capability. The adoption measures are smaller: £30 million for local AI partnerships with three Yorkshire mayoral authorities, a £60 million AI Pathfinder Fund and £23 million for the AI Economics Institute, according to techUK’s analysis of the Budget.

techUK’s own view is that £30 million split across several authorities will not move adoption rates by itself, although it could produce pilots that other regions copy. Based on the measures published, there is no broad grant or tax credit of the kind the British Chambers of Commerce called for in March to help SMEs invest in AI and training. For most small businesses, closing the gap will be self-funded for now.

How To Close The Gap In Your Own Business

The firms that get results tend to follow a similar sequence. None of it requires a large budget.

1. Start with a process, not a tool

Pick one high-volume, repetitive process that costs real hours every week, such as handling new enquiries, chasing invoices, onboarding clients or compiling a weekly report. A specific process gives you something to measure. A general instruction to “use more AI” does not.

2. Map the process before you automate it

Write down every step, who does it and where information gets copied from one system to another. Most of the waste in small businesses sits in those handoffs, and automating a broken process only produces the same mistakes faster.

3. Connect your systems first

AI works better when it has context. Linking your CRM, accounting software and project tools through an integration platform means an AI step can read real customer and job data instead of whatever someone pastes into a chat box. If you are choosing a platform, our comparison of Make and Zapier covers the main options.

4. Measure a baseline

Record how long the process takes, how often it goes wrong and how quickly it finishes before you change anything. Check again at 30, 60 and 90 days. Without a baseline, you are relying on impressions, and impressions are exactly what the national surveys are picking up.

5. Train people and set review rules

Decide which outputs need a human check and which do not, and show staff what good use looks like on real work. Training was the clearest dividing line in the Zapier research, and skills were the biggest barrier in the government survey.

6. Get help when the stakes justify it

For core processes, a few days with an experienced automation or AI consulting firm can save months of trial and error. The aim is not to outsource the thinking, but to get the design right before the business depends on it.

Frequently Asked Questions

How many UK small businesses use AI?

The government’s 2026 survey found that 27% of businesses with 5 to 9 employees and 34% of those with 10 to 49 employees use at least one AI technology. Other surveys report higher figures because they use different samples and definitions, but all show adoption rising quickly.

Is AI actually making UK businesses more productive?

For about half of adopters, yes, but the gains are mostly modest. Half of businesses using AI reported higher employee productivity in 2026, down from 56% in 2025. The small group of firms that have built AI into their core processes report much stronger results.

What is a good first AI project for a small business?

A repetitive admin process that already runs through software you use, such as lead handling, invoicing or reporting. These are easy to measure and easy to connect to existing systems. Our guide to AI agents for small business lists twelve common starting points.

Is there government funding for AI adoption?

Support is limited at the moment. The Autumn Budget funded local AI partnerships starting in three Yorkshire mayoral authorities, along with a new AI Pathfinder Fund. Free AI skills courses are also available through the government’s AI skills programme.

The Short Version

UK businesses have doubled their use of AI in a year, but most are using free tools on everyday admin and few have changed how their work is done. Reported productivity gains have shrunk as adoption has spread. The firms getting real results are the ones that pick a specific process, connect their systems, measure the outcome and train their people. That is less about buying AI and more about running the business well, and it is available to any firm willing to do the groundwork.