A FIS survey of 100 UK finance directors finds 65% sometimes invest without a full view of liquidity, while many struggle to measure what AI delivers.
According to new research from FIS, 90% of UK finance leaders have a clear business case for using AI in the management of cash flow and liquidity. However, only 71% are actually able to measure the value delivered by their AI projects.
The fintech company commissioned research firm Opinium to survey 100 senior finance directors at UK-based enterprises with an annual turnover of £250m or more. The fieldwork was conducted between June 25 and July 4, 2026.
Respondents have little doubt about the technology itself. Approximately 89% state they understand where AI can add the most value within their field, and 91% claim that their teams possess the right skills to deploy AI effectively.
The remaining 29% – who cannot demonstrate the returns from their AI initiatives – form the core focus of the report; according to FIS, this points to an issue with evidence, rather than a lack of confidence in the technology.
A similar distance between responsibility and oversight is evident in how companies monitor cash flows. Two-thirds (65%) of finance leaders indicate investment decisions are sometimes made without a complete picture of available liquidity.
This is in contrast to the near-unanimous view (97% of respondents) the finance department is increasingly expected to be accountable for cash flow and liquidity outcomes – even when the underlying data, systems, or processes fall outside their direct control.
A significant portion of a finance teams’ workweek is consumed by preparatory tasks. On average, finance teams spend 16.2 hours per week gathering, reconciling, and validating data on cash positions before it reaches senior management. This leaves less time for the analysis expected of them.
FIS: obstacles to scaling AI
Measuring outcomes is not the only stumbling block. Among those surveyed, 72% indicate concerns regarding accuracy, explainability, or governance are hindering the adoption of AI, while 76% point to risks related to data security.
James Mitchell, Global Head of the Office of the CFO at FIS, notes the ability to measure the value of AI “is not keeping pace with deployment.”
“If you cannot clearly demonstrate what value AI is delivering, it becomes very difficult to build the internal confidence needed to support a move beyond pilots and truly scale,” Mitchell said. “AI is only as reliable as the information it works with, and in many finance functions, that information is still being assembled manually from disconnected systems.”
Fragmented structures
Financial leaders in the United Kingdom manage an average of 14 legal entities, 11 banking relationships, and 10 core systems for finance, treasury, ERP, or payment processing. This fragmentation makes it difficult to maintain a consolidated, real-time view of the liquidity position, the report notes.
Only 78% indicate their systems are sufficiently interconnected to support AI applications, and 79% state they have access to the data required for AI initiatives without needing significant manual effort.
Consequently, about one in five organisations lacks the foundation necessary for AI to function using reliable liquidity data.
When asked what would make the biggest difference, financial leaders most frequently cite (46%) the further automation of reporting and reconciliations.
An integrated, real-time overview of the liquidity position across all entities and accounts – as well as better system integration – are each mentioned by 43% of respondents, while improved interoperability between systems is cited by 42%.
According to FIS’ Mitchell, the core of the problem lies in the quality of the data feeding both people and models.
“Finance leaders don’t lack conviction in AI. They often lack the evidence to prove it, and you cannot safely scale what you cannot trust and measure,” he said.
“That measurement gap is really a data quality gap. When a cash position is assembled from disconnected banks, entities and systems, with any manual process involved, neither the finance team nor an AI model can fully utilise it until that data has been rationalised.”
He adds decision-making processes can cause delays, and trust in AI models may decline if companies deploy the technology before this preparatory work has been completed.
“We expect the organisations that pull ahead to be the ones that address their data foundation and interoperability issues first,” Mitchell said.
