InsightsAI Insights
AI Insights

Deeper, not wider: the £42.6bn case for doing more with the AI you already use

For most UK small businesses, the AI question has quietly changed. It is no longer "should we use it?" Most already do. The question now is why some firms are seeing real revenue from it while others are mostly saving a bit of time on emails.

By StJohn Krog, Founder & MD

AI Insights21 October 2026

Three pieces of research published this month point to the same answer. The money is not in using AI more widely. It is in using it more deeply.

The headline number, and the catch

On 19 October the Social Market Foundation (SMF) published Byte-sized growth, a report sponsored by Intuit and based on a survey of 1,015 SME leaders combined with ONS data on local economies. Its central estimate is that wider AI adoption could add around £42.6bn to SME revenues across Great Britain, within a range of £38.1bn to £47.3bn.

The adoption picture in that survey is already high:

  • 79% of SME leaders said their business uses AI in some capacity: 92% of medium-sized firms, 86% of small firms and 60% of micro businesses.
  • 87% of AI users reported a positive impact, against 1% reporting a negative one.
  • The most common benefits were time savings (88%) and better quality of work (79%). 55% reported higher revenue, and current users report an average revenue uplift of about 2.8%.

The catch is in the detail. The SMF found that advanced users report significantly better results than basic users, and that specialist tools show a stronger, more consistent link with positive outcomes than general chatbots. Process-heavy areas (HR, sales, supply chain and IT) looked like the highest-value places to apply it.

It is also worth being clear about what the £42.6bn is: a modelled scenario that assumes the firms planning to adopt AI actually do, with later adopters gaining a bit less. It is a sense of scale, not a forecast.

Most firms are still on the bottom rungs

If the gains come from depth, how deep are UK businesses actually going? Not very.

The Reality Department's Business Reality Index (500 senior decision-makers at UK organisations with 50+ employees, published 8 October) breaks AI use into rungs:

  • 32% are not using AI.
  • 25% say the only use is individual employees using it informally.
  • 16% are piloting it in specific functions.
  • 15% have deployed it in one or two core processes, and 9% across multiple core functions.
  • 2% have AI agents running with limited oversight.

So only around a quarter (26%) have AI in a core process at all. AI and automation was still the most-named investment priority (22%), roughly twice the share naming cyber security and resilience (11%).

The ONS gives the economy-wide view. Its Business Insights survey (published 8 October) found 30% of all UK businesses used at least one AI technology in September 2026, up from 9% in September 2023. That rises to 51% for businesses with 250+ employees and falls to 29% for those with fewer than 10. The most common uses among firms with 10+ staff were text generation (21%) and visual content creation (19%). Those are useful, but they are mostly the "basic user" end of the SMF's spectrum.

(The surveys measure different things, which is why 30%, 68% and 79% can all be true. ONS asks about specific technologies across all firms; the others ask leaders about any AI use, including informal.)

Leaders and staff are not on the same page

Microsoft's UK research, released on 20 October ahead of its Frontier Tour in London, adds a people angle. In a study of 5,000 UK employees and 1,000 senior decision-makers:

  • About 90% of business leaders say they have a clear AI strategy.
  • Only 41% of employees are confident the UK can become a world-leading AI economy within five years, against 84% of leaders.
  • Among employees already using AI at work, 71% report a positive impact on their work or career, but 35% worry it has lowered their own job security.
  • Skills and training are a top-three priority for both groups.

That gap matters for depth. You cannot move AI from someone's browser tab into a core process without the people who run that process trusting it, understanding it and helping to shape it.

How to go from "we use ChatGPT" to "AI runs part of how we work"

  1. Pick one process, not one tool. Start where the SMF saw most value: something repetitive and rules-heavy in sales, HR, supply chain or IT support. Quote preparation, order exceptions, candidate screening and ticket triage are all good candidates.
  2. Write down the current process first. Who does what, which systems hold the data and where the delays are. If you cannot describe it, AI will only automate the confusion.
  3. Prefer tools that live where the work happens. The SMF's finding on specialist tools makes sense: AI built into your CRM, helpdesk, ecommerce platform or CMS has the context a general chatbot lacks. Check what your existing software already offers before buying something new.
  4. Connect it to your data, carefully. The value comes when AI can read real orders, tickets or records, so set permissions, decide what it can change and keep a human approval step until results are proven. Data security was the most-cited barrier in the SMF survey (33%), and for good reason.
  5. Bring the team in early. Ask the people doing the work what they would hand over first. Give them training time, not just a login. It closes the confidence gap Microsoft found.
  6. Measure in money and hours. Record a baseline (time per task, error rate, conversion, cash collected) and compare after a month. "It feels quicker" will not survive a budget review.
  7. Turn informal use into a policy. If staff are already using AI on their own, agree which tools are approved, what data must never go in and who to ask. That moves you up a rung without spending anything.

None of this needs a big programme. One process done well, measured and then repeated is how firms move from the 74% to the 26%.

Why this matters for Clarity Growth

Clarity Growth helps UK SMEs and scale-ups do exactly this: choose the right process to start with, map how it really works, and connect AI into the systems they already run, whether that is a CRM, a Shopify store, a Statamic website or a project management tool like Jira. We design the integrations, permissions and human checkpoints so AI can work with live data safely, and we set up the measurement that shows whether it is paying back. If your business is using AI but not yet seeing it in the numbers, we can help you pick the one process where going deeper will pay off first.

Sources

Related services

AI & automation

Automation

Technology consultancy