SMEs should frame AI as a productivity driver rather than a strategy

Artificial intelligence (AI) is widely positioned as a growth and innovation accelerator. But for most SMEs, AI is better framed as a productivity driver rather than a strategy. 

AI adoption levels are rising at an uneven pace across different sectors. At the national level, more than one-third of UK SMEs (35%) are actively using AI (2024: 25%), research by the British Chambers of Commerce (BCC) shows. A further 24% of UK SMEs plan to adopt AI, and 33% have no plans (down from 43%). But within this rising trend, only around 11% of SMEs report using AI to a “great extent” across their operations.

AI adoption is heavily concentrated in the finance, law and marketing sectors – utilised in knowledge-intensive tasks such as document drafting, data analysis, interpreting outputs and supporting business decision-making. By contrast, adoption is lagging in customer-facing and operational, labour-intensive sectors, such as retail and construction. In the retail sector, a BCC poll of 500 UK employees found that 72% never use AI at work, which is attributed to poor training. Nine out of 10 (89%) retail employees surveyed reported they had received no AI training. 

In a Europe-wide survey, more than half of respondents (51%) were concerned that their business was not utilising AI effectively. Employee concerns were related to an inadequate in in-house technical proficiency and a lack of trust in AI-generated outputs. These challenges were replicated at the sector level. In real estate, the proportion of firms running AI pilots has soared in the past three years – from 5% to 92%, with only 5% reporting that most objectives have been achieved. This suggests that either the lag time between investment and measurable success is still long, or that AI experimentation often fails to translate into productivity gains as adopters get to grips with the technology. 

SMEs cite cost uncertainty, training requirements, and the need to create appropriate governance frameworks as factors that continue to slow AI investment. In this context, SME boards and directors must decide – with limited time, capital and internal resource bandwidth – where AI investment should be focused to deliver measurable productivity gains. This discipline is good practice across all discretionary technology investment.

Starting points for SMEs

Integration should begin with specific business problems and clearly defined inefficiencies, rather than broad transformation agendas. Back-office internal functions are a common starting point, such as document preparation, work scheduling, invoicing, reporting, and internal communications. 

The shift from internal to external functions requires greater consideration and confidence in AI tool efficacy, as they involve client interaction. Early-stage external AI functions include responding to first-line enquiries, frequently asked questions and drafting marketing content. These contexts exemplify how AI works as a productivity layer that reduces human effort and reclaims time while retaining human decision-making and organisational judgement. These low-risk applications are also relatively easy to reverse if they do not prove sufficient productivity gains. Above all, accountability remains with people, not systems.

Benchmarking

Without disciplined measurement and benchmarking, AI investment risks becoming another fixed cost. AI tools are often trialled informally by individual employees or teams, but results are not benchmarked or consistently captured at the organisational level. Over time, this can fragment processes and create governance blind spots. This underlines the importance of AI initiatives to be treated as provisional projects until proven otherwise. Scaling up pilot schemes should be contingent on demonstrable gains in productivity, efficiency or decision quality.

Time reclaimed should be treated as a primary productivity metric as it frees up staff to focus on higher-value activities such as client relationships, sales, value creation, and operational oversight (which AI governance requires more of). However, objective benchmarking is essential. Reporting time savings must be offset against staff training and cost, system oversight and quality control.

Governance 

AI’s promised benefits need to be balanced against real-world risks – including operational, financial and sector-specific ethical and legal risks. These risks become materially harder to manage once AI is embedded across corporate services and systems. Scaled AI integration requires upfront investment, including specialist labour to oversee workflow redesigns, and increased oversight responsibilities. These organisational changes create new governance risks. These include: 

• Data governance: Without deliberate controls, management may lose visibility over data sources, access permissions, and exposure of sensitive or personal information. AI outputs are underpinned by large multi-modal models (LMMs), which demand vast quantities of structured and unstructured data. This places enormous pressure on digital infrastructure, governance protocols, and cybersecurity systems, and increases regulatory compliance responsibilities, which may carry ‘non-compliance’ fines. Risk frameworks grounded in ‘zero trust’ security models that comply with privacy regulations (i.e., GDPR) are increasingly being adopted. Fundamentally, SMEs face a permanent increase in governance oversight, security audits and cybersecurity capabilities to manage AI. 
• Reputational and insurance exposure: AI output inaccuracy is a persistent weakness. For example, where AI contributes to financial, legal or health-related judgements and results in a detrimental outcome, the reputational and litigation risks are high. As a result, AI-driven decisions must be auditable, particularly in regulated sectors. 
• Vendor lock-in: An unintended dependency on AI software providers can create lock-in risk that increases costs and reduces operational flexibility over time. Over-commitment ahead of establishing a governance framework risks amplifying operational risk.

Many SMEs lack the internal capacity to design or oversee complex AI strategies and are less able to absorb potentially costly setbacks. This reinforces a preference for limited and staged adoption. For SMEs, AI is best deployed as a controlled productivity layer rather than the strategic backbone for wholesale business transformation. SMEs that benefit most will be those that treat AI as tightly aligned to measurable operational outcomes. 

SME’s AI adoption is inevitable. But those who succeed will not necessarily be the fastest adopters – they will be those who manage complexity with discipline, balancing innovation with oversight and productivity gains with strong governance controls. BTG helps SMEs navigate these risks and build the operational resilience to adopt AI responsibly. Our expertise in forensic risk assessment, digital infrastructure stress testing and operational turnaround enables clients to move forward at a pace aligned with their resources, controls and priorities.