The True Cost of
Getting AI Wrong

When UK e-commerce retailers invested £120 million in AI in 2024, they expected transformation. Better customer experiences, improved efficiency, and a genuine competitive edge.

Instead, 77% now admit their AI initiatives are falling short. That’s approximately £92 million of investment at risk, spent on technology that’s simply not delivering the expected value.

The financial figure is serious.

But it’s only part of the story.

The costs that do not appear on the balance sheet

Wasted investment is the visible consequence. The harder costs go deeper.

When a project fails, internal confidence drops. “We tried AI, and it did not work” becomes the default response. Future proposals face higher barriers, and progress slows when momentum matters most.

There’s also the real cost of staff time. Senior leaders, IT teams, and business units spend months on an initiative that produces nothing. That time is gone, and the opportunity cost is high.

Then there are regulatory risks. An AI system launched without adequate governance can process data inappropriately or make decisions it cannot explain. Investigations, remediation, and reputational damage can exceed the original spend.

Why failures follow the same patterns

UK retailers identified the AI applications most likely to underperform:

• Chatbots at 29%
• Data analysis tools at 27%, and
• AI-driven marketing at 23%.

These are not failures of the technology. Modern AI works when the conditions are right. It fails when organisations deploy it without the foundations it needs: clean data, a clear use case, realistic expectations, and strong governance.

The same pattern repeats across other sectors.

• A manufacturer invests in predictive maintenance AI, only to discover their sensor data is too inconsistent for a model to learn from.
• A government department launches a pilot but stalls when legacy systems cannot provide the required data inputs.

In every case, the technology was not the problem.

The foundation was.

What changes when you get the foundation right

Gartner found that only 48 percent of AI projects reach production, taking an average of eight months from prototype to deployment. For the remaining projects, the investment delivers little or no value.

Organisations that succeed tend to share one characteristic, they:

• Understood their readiness before they began
• Knew which data gaps to close
• Had realistic timelines
• Built governance alongside the technology

The difference isn’t talent or budget.

It’s approach.

Where a clear strategy makes the difference

Nexer Insight’s AI Strategy and Roadmap engagement isn’t a long consulting exercise. It’s a focused conversation about where you are, where you want to be, and what it takes to get there before major spend is committed.

It can reveal data issues that would take months to discover mid-project. It can set realistic expectations for ROI and timelines. It can identify governance gaps before they become compliance problems.

The cost of not having that conversation is visible in the statistics.

The cost of having it isn’t.

Get your AI strategy right from the start: https://nexergroup.com/uk/data-ai-advisory/ai-strategy-roadmap/