“We do not have the time.”
Smaller firms rarely have spare capacity, and the constraint is a real one. Implementation does not begin with a programme. It begins with one process, chosen because it already consumes more time than it should, and the first result is time returned.
The evidence. Among the organisations attributing the most profit to AI, nearly three-quarters had redesigned workflows around it. In our experience that begins with a small number of core processes, not with everything at once.
“AI is not relevant to a business like ours.”
This is the most common view among smaller firms: 55% of UK SMEs surveyed in May 2026 could not see the relevance, and in a government survey 71% of businesses cited having no identified use for it. The useful question is not about AI. It is whether any process in the business is slow, prone to error or dependent on one person. Every business has such processes.
The evidence. Among small firms that had adopted AI in the previous 12 months, 59% reported a productivity gain.
“We tried it and saw no result.”
That is the most common outcome. It is usually the result of adoption without implementation: a tool was introduced and the process around it stayed the same.
The evidence. Among the organisations attributing the most profit to AI, nearly three-quarters had redesigned workflows around it, against about one in four of all others.
“The output cannot be trusted.”
The concern is well founded. It is the most common concern among small firms, cited by 54%, and on a task outside the competence of the AI, people using it were 19 percentage points less likely to reach the correct answer. That is an argument for selecting tasks carefully and building verification into the process. It is not an argument for abstaining.
The evidence. In our own analysis of more than 750 communications between business owners, those deriving the most value were those who consistently verified what the AI returned.
“We will wait until the technology settles.”
The tools are changing quickly, and a tool chosen today may well be replaced. The disciplines of implementation are disciplines of management. A process that has been mapped, measured and assigned an owner remains an asset whichever tool is later built into it.
The evidence. The gap in output per worker between the leading firms and the median was 2.9 times before 2008 and 3.5 times in 2023, and has sat between 3.4 and 3.5 since 2020. It has not closed on its own.
The Implementation Gap sets out the evidence behind each of these answers, and a first step for each stage of implementation.
Sources
- Digital and AI adoption among UK SMEs, Enterprise Nation with Strand Partners, June 2026.
- AI Adoption Research, Department for Science, Innovation and Technology, January 2026.
- The Confidence Code, Federation of Small Businesses, July 2026.
- Dell'Acqua et al., "Navigating the Jagged Technological Frontier", Organization Science, March 2026.
- Office for National Statistics, Trends in UK business dynamism and productivity: 2025, December 2025.
Full references are given in the report.