AI ROI isn't obvious, and that's the point

There is a comfortable story told about AI hesitation in British business. It goes like this: the technology is clearly valuable, the evidence is everywhere, and the only thing holding companies back is nervousness or resistance to change. If that were true, the fix would be simple. Run an internal workshop, reassure the sceptics, and press on.

The data does not support that story. Recent research cited by The Independent found that around a quarter of small and medium-sized enterprises identify uncertainty around AI return on investment as a primary adoption barrier. That is not fear of change. That is a reasonable request for evidence.

For SME leaders and IT decision-makers, this distinction matters enormously. Misdiagnosing the problem leads to the wrong remedies.

Why the 'just get on with it' narrative persists

The belief that AI value is self-evident is understandable. Productivity headlines are frequent and bold. Major consultancies publish surveys showing significant efficiency gains. Technology vendors cite impressive figures. If everyone else seems to be benefiting, reluctance can look like a cultural problem rather than a rational one.

There is also a grain of truth in it. Some AI applications genuinely do deliver quick, visible wins. Drafting routine emails faster or summarising long documents are tasks where the improvement is immediate and obvious. These examples travel well in case studies and conference talks.

The difficulty is that these headline use cases are not representative of how AI creates commercial value at scale. The easy wins are real. Sustained, measurable return on a meaningful investment is a harder argument to make, and most vendors are not helping businesses make it rigorously.

What the evidence actually says about UK adoption barriers

The picture emerging from UK research in 2026 is more nuanced than the 'change resistance' narrative suggests.

A YouGov business sentiment tracker reports that 39% of UK businesses cite accuracy and reliability concerns as barriers to AI adoption. Data privacy and security concerns follow at 29%. Ethics is named by 22%. A lack of internal skills is cited by 20%. Notably, 12% of UK businesses say they do not see AI's value at all.

These are not the responses of organisations gripped by irrational technophobia. They are organisations asking sensible questions that vendors and consultants have often failed to answer clearly.

The SME-specific data is sharper still. Research cited by The Independent places ROI uncertainty alongside trust and data privacy as the leading barriers for smaller businesses. For IT decision-makers in particular, the question is not whether AI is interesting. It is whether a given investment can be justified with numbers that will stand up to scrutiny from the finance director or the board.

That is a completely reasonable position. It is also, notably, the same standard applied to every other technology investment.

What believing the myth costs in practice

When organisations treat AI hesitation as a change management problem, they respond with the wrong tools. They invest in internal communications about AI's potential. They run awareness sessions. They share case studies from other sectors.

None of that answers the underlying commercial question.

The result is a predictable cycle. A pilot runs without defined success criteria. Results are described qualitatively rather than measured. Leaders cannot determine whether the investment justified itself. Confidence does not build. The next proposal for AI investment faces the same scepticism, now compounded by a previous initiative that produced no clear verdict.

For SME leaders, this cycle is particularly costly. Resources are tighter. Tolerance for ambiguous returns is lower. A £30,000 to £50,000 AI initiative that cannot demonstrate its impact is not just a missed opportunity. It actively makes the next initiative harder to approve.

IT decision-makers face a related challenge. They are often tasked with evaluating and recommending AI investments, but without a clear framework for measuring outcomes, they cannot build a compelling internal case. The result is either paralysis or projects that move forward on enthusiasm rather than evidence, creating the accountability gap that ROI uncertainty describes.

Three practical ways to build a measurable AI investment case

The commercial standard for AI investment is identical to the standard for any other capital or operational expenditure. That framing alone removes significant confusion.

1. Start with the business problem, not the technology. Identify a specific process with a known cost or a known risk. Customer query response times, manual data entry hours per week, error rates in routine document processing. These are quantifiable. They become your baseline.

2. Define success before you deploy. Agree in advance what a good outcome looks like. A 30% reduction in query handling time. A 20% reduction in data entry hours. A measurable decrease in error rates over a 90-day pilot. These thresholds let you reach a clear verdict rather than a narrative.

3. Run a contained pilot with clean measurement. Pick one or two use cases, not five. Apply AI to the defined process, hold other variables constant where possible, and measure against your baseline at the end of the pilot period. The output is a number, not an impression.

This approach is not new. It is how any competent operations or finance team evaluates any significant process change. The only thing AI adds is the need to account for variability in output quality, which makes the accuracy and reliability concerns in the YouGov data entirely rational rather than obstructive.

Regulatory reality: compliance is part of the ROI calculation

AI-driven productivity and financial gains cannot be assessed in isolation from compliance requirements. UK data protection rules apply to how AI systems process personal data. Where businesses operate in regulated sectors, FCA or sector-specific guidance adds further obligations.

This is relevant to the ROI calculation in a direct way. A process that saves 10 hours per week but introduces material compliance risk does not have a positive return. The full cost of an AI initiative includes the cost of ensuring lawful processing, appropriate human oversight, and transparency where required.

Building this into the business case from the outset is not defensive. It is accurate. It also addresses the data privacy concerns that the YouGov tracker identifies as the second most common barrier, sitting just below accuracy and reliability at 29%.

For IT decision-makers in particular, framing compliance as part of the commercial model rather than a constraint on it tends to accelerate internal approval rather than slow it down.

A realistic approach to closing the ROI gap

The confidence gap in UK AI adoption is not closing quickly. The organisations making progress are those that have stopped treating AI as a category decision and started treating it as a portfolio of specific, measurable investments.

The honest caveat here is that baselining is harder than it sounds. Many SMEs do not have clean data on process performance before an AI intervention. Customer query handling times may not be systematically logged. Manual data entry hours may not be tracked. Building the measurement infrastructure alongside the AI deployment adds time and cost that straightforward ROI projections often ignore.

This is worth acknowledging plainly, because underestimating it is one of the reasons AI projects fail to produce the evidence their sponsors need.

For businesses that want a structured route through this, gecco's AI Agents service addresses exactly this gap. ROI uncertainty most often stems from deploying AI without clear process mapping or quality gates, then finding it impossible to isolate the impact. Structuring AI as autonomous agents with measurable handoffs converts vague value into auditable outcomes, giving SME leaders and IT decision-makers the numbers they need to make and sustain the commercial case.

Your next step if ROI uncertainty is the real barrier

If your organisation has been treating AI hesitation as a people and culture problem, but the real question is commercial, the AI Readiness survey is the right starting point. It surfaces where you are in the adoption journey and, critically, where the measurement gaps sit.

Take the AI Readiness survey. You will get access to 65+ free resources and a custom AI Readiness report. We then offer a free 45-minute AI Readiness call to walk through your results.

If you want support in structuring AI investments with clear process mapping, quality gates, and auditable outcomes, that is exactly what gecco's AI Agents service is built to deliver.

Website · LinkedIn · Case Studies · Newsletter