AI value stalls on cost, skills and data

Three barriers keep coming up whenever UK business decision makers and SME managers explain why AI has not yet delivered what they hoped. Cost is too high. Data is not ready. The team does not have the skills. These feel like firm reasons to wait. A Lloyds Banking Group survey published in August 2026 suggests otherwise. The same businesses citing these barriers are already seeing AI create new roles and improve productivity. The barriers are real, but they are execution gaps, not permanent blockers. This article examines what the evidence actually shows and what a realistic path forward looks like.

Why these three barriers feel so convincing

The Lloyds survey found that 18% of UK firms cite cost as a barrier to extracting full AI value. Data quality and access to skills were each cited by 17%. Those numbers are meaningful, and the concerns behind them are legitimate.

Cost concern usually comes from watching large enterprises announce AI programmes worth tens of millions. If that is your reference point, AI looks like something only well-capitalised businesses can afford. Data concern comes from a reasonable instinct: AI is only as good as the information you feed it. If your records are incomplete or inconsistently formatted, why would AI help? Skills concern reflects the genuine shortage of AI specialists in the UK labour market. If every implementation seems to need a data scientist or a machine learning engineer, smaller businesses are right to feel excluded.

All three concerns share the same root: the mental model is wrong. Most of these firms are comparing their situation to a full-stack enterprise AI programme. That is not the relevant comparison.

What the evidence actually shows

The Lloyds survey also found that more than half of UK businesses say AI has already created new jobs. These are not all large firms with dedicated AI budgets. The evidence reframes hesitation not as a question of commercial readiness, but as an execution and capability question.

On cost: the firms extracting value are not necessarily spending more. They are spending more precisely. Narrowly scoped use cases, targeting one workflow or one department, cost a fraction of an enterprise-wide rollout. The return is also measurable sooner, which makes the investment easier to justify internally.

On data quality: the common mistake is assuming all data must be clean before any AI work begins. In practice, a single well-defined use case requires clean data for one specific workflow. That is a manageable data quality task. It is not a full data infrastructure overhaul. Starting narrow means the data preparation work is proportionate.

On skills: the UK AI skills gap is real, but it does not need to be solved company-wide before any value is realised. Targeted training for the team members who will use AI in their day-to-day work closes the practical gap. External implementation support can handle the initial setup. The goal is building enough internal confidence to run a controlled deployment, not producing in-house data scientists.

Regulatory context matters here too. Poor data quality and weak internal skills make GDPR-compliant AI deployment harder. If team members cannot explain or govern AI outputs, that is a compliance risk. Structured skills development is not just about productivity. It is also about being able to account for how AI decisions are made.

What believing the myth costs in practice

For business decision makers and SME managers, the cost of inaction is not neutral. While the 18% citing cost as a barrier wait for a better moment, competitors in the same sector are running pilots and building capability. The skills gap compounds over time. Staff who are not using AI tools are not building familiarity. That familiarity gap becomes harder to close as AI becomes more embedded in standard workflows.

The data quality concern, left unaddressed, also compounds. Firms that wait for a perfect data environment before starting often find the environment never becomes perfect. Data quality improves through use, governance, and iteration, not through indefinite preparation.

The practical cost is not the money spent on AI. It is the months of productivity and capability development that pass while waiting for conditions that never quite arrive.

Three things to try before committing to a full programme

The evidence points to a consistent pattern among firms that successfully close these gaps. They do not start with a strategy. They start with a specific problem.

1. Identify one repetitive workflow that currently takes significant time. Do not try to automate everything. Pick the single task where time savings would be most immediately visible to the team.

2. Audit only the data that workflow needs. Not all your data. Just the records, documents, or inputs that this specific task relies on. That audit is usually a matter of days, not months.

3. Train the team members who will use the output, not everyone. Targeted skill-building for the people directly involved in that workflow builds confidence quickly and keeps the investment proportionate.

This approach does not require a large budget. It does not require a data scientist. It does not require the organisation to be AI-ready in some comprehensive sense. It requires a clear scope and a willingness to start small.

A realistic approach for SME managers and decision makers

The three barriers are not myths in the sense of being invented. They are real constraints that become blockers when the mental model is wrong. The intervention is not to dismiss cost, data, or skills as concerns. It is to scope the work so those concerns are proportionate to the actual task.

For SMEs specifically, the false choice is between doing nothing and hiring an AI team. Neither is necessary. A structured onboarding engagement spreads the initial cost, builds internal capability, and delivers early wins that make further investment easier to justify. Firms citing cost as a barrier may be confusing investment in capability with ongoing platform spend. Those are different things with different timelines and different returns.

One option for SME leaders who want structured support is gecco's Quick-Start engagement, which is a four-week onboarding programme designed to guide internal teams through practical AI implementation with structured frameworks and training, without requiring existing in-house AI expertise.

Honestly, any approach that starts narrow, prioritises data readiness for one workflow, and includes team training will outperform a broad programme that tries to address everything at once. The evidence is consistent on this point. Scope is the variable that separates firms extracting value from firms still waiting.

Where to start if you are weighing up the cost

If your team is sitting with cost, data, or skills as reasons to wait, that is exactly the situation the AI Readiness survey is built to surface and work through.

Take the AI Readiness survey. You will get access to 65+ free resources and a custom AI Readiness report. We then offer a free 30-minute showcase to walk through your results.

If you want structured help turning that readiness into a scoped first deployment, gecco's Quick-Start programme is built for business decision makers who need early wins without a large upfront commitment.

---

Website · LinkedIn · Case Studies · Newsletter