UK businesses lose a quarter of AI budgets to complexity

Freshworks has published research indicating that UK businesses lose an average of 24% of their AI budgets to complexity before seeing any return. The survey also found that 83% of UK IT leaders say AI outputs introduce noise, errors or rework.

It estimates that wasted spending totals GBP £11.7 billion a year in the UK. The findings are based on a survey of 12,021 IT decision-makers across six countries, including more than 9,000 respondents from mid-market organisations.

The report argues that mid-market companies face a heavier burden than larger enterprises because they have fewer staff and less budget flexibility to deal with integration issues, governance demands and flawed outputs from multiple AI tools.

Across the global sample, 86% of mid-market IT leaders said AI complexity had increased their team's workload rather than reduced it. Eight in 10 said AI outputs were creating noise, errors or rework every day, while respondents reported spending an average of 25% of AI-related time fixing poor outputs instead of focusing on other work.

That burden is rising as businesses continue to increase AI spending. Nearly 89% of mid-market organisations surveyed said they planned to raise AI investment over the next 12 to 24 months. Yet only 15% said AI had been integrated across core business operations, while 36% said projects remained at the pilot stage.

ROI pressure

The research points to a gap between executive expectations and the practical realities of deployment. While 72% of mid-market executives expect AI investments to deliver a return within eight months, 55% of organisations said deployment alone takes six to 12 months before meaningful returns can begin.

Freshworks identified system integration complexity, shortages of skilled staff and extensive configuration requirements as the main reasons pilot projects fail to expand into broader programmes. These factors create a risk that projects are judged too early and cut before they have a chance to produce measurable results.

The findings also suggest that the spread of multiple AI products within a single business is adding to the problem. Mid-market organisations reported using an average of 4.2 AI tools, and 10% said they were running seven or more. Yet only 33% said they had a formal, consistently applied AI governance framework.

That combination of tool sprawl and weak oversight leaves smaller IT teams dealing with duplicated work, inconsistent outputs and extra checks. For mid-sized organisations, the issue is not only the direct cost of software but also the staff time needed to make systems work together and verify the quality of AI-generated output.

Srinivasan Raghavan, Chief Product Officer at Freshworks, said the market was shifting towards simpler deployments and faster results. "Mid-market IT leaders don't have time for AI that takes months to deliver value. They need AI that works inside the business they already run and shows value fast," he said.

The study suggests buying preferences are changing as a result. A third of mid-market IT leaders said workflow integration would be their top priority over the next two to three years, 90% said they preferred built-in workflows over heavy configuration, and 54% said they were buying AI tools rather than building them in-house.

Mid-market caution

Doug Farren, Executive Director of the National Centre for the Middle Market, said that pattern reflected a more cautious adoption path among medium-sized companies. "Middle market businesses tend not to be early innovators and often lag in realizing full-scale implementation benefits until they are confident of ROI. Until then, smaller pilots and tests are often used to prove feasibility," he said.

The data adds to a broader debate over whether AI is delivering measurable productivity gains for businesses outside the largest corporate groups. Large enterprises often have the budget to add compliance teams, specialist engineers and consultants around new systems. Mid-market companies are less able to absorb those costs, even as pressure from senior management to adopt AI remains high.

Freshworks surveyed IT decision-makers at director level and above in the UK, US, Germany, France, Singapore and India. It defined mid-market organisations as those with up to 5,000 employees.

The figures indicate that, for many UK businesses, the financial challenge lies less in buying AI than in making it usable at scale, with almost a quarter of spending disappearing before any return is visible.