Closing the gap requires leadership discipline, not more tools
AI investment in financial services is accelerating. But for many firms, the ability to demonstrate what it’s actually delivered remains frustratingly unclear. This article explores why the gap between AI activity and measurable business value exists, what causes it, and what leadership teams need to put in place to fix the issue.
Key Takeaways
- Many financial services firms are investing in AI but struggling to demonstrate returns at a business level
- The problem is rarely the technology but the absence of clear problem definition, success metrics, and ownership
- Firms running AI in production are seeing revenue and productivity gains, but only when programmes are set up correctly
- High-ROI use cases in financial services include fraud detection, compliance automation, and customer service
Can you explain the ROI your AI has delivered?
For many financial services leaders, that question is harder to answer than it should be. It’s normally because a connection between AI activity and measurable business value hasn’t been clearly established.
Teams are running pilots. Tools are being deployed. Capability is growing. And yet in boardrooms across the sector, the same question keeps coming up:
“What is all this actually giving us?”
It’s more a leadership problem than an issue with technology, and it’s more common than most organisations want to admit.
The teams we speak to tell a consistent story. Investment feels significant, momentum feels real, and yet when pressed on what AI has delivered in business terms, the evidence is thin. Programmes that looked promising have stalled. Value that felt imminent has proved hard to quantify.
What does the data say about AI ROI in financial services?
According to Deloitte’s 2025 survey of 1,854 executives, 85% of organisations increased their AI investment, and 91% plan to increase it again this year. However, most also reported a lag of two to four years to see satisfactory returns on a typical AI use case; far longer than most leadership teams are planning for. A Boston Consulting Group survey found that fewer than half of finance executives can quantify what their AI investment has delivered at all.
The productivity picture is equally telling. Individual gains are real. People are saving time, moving faster, and reporting higher output. But those gains aren’t translating into organisational performance. Atlassian’s AI Collaboration Index found that 96% of companies don’t see AI ROI and that focusing on personal productivity over innovation could cost the Fortune 500 $98 billion annually in lost returns on their AI investments.
The gap isn’t between AI and value. It’s between how most organisations are measuring success and what actually drives enterprise-level return.
What are the gaps holding your AI back?
In our experience working with financial services organisations, the same three gaps appear repeatedly in programmes that are struggling to demonstrate value.
- 1. No clear problem definition
Many AI initiatives begin with a solution looking for a problem. Technology is acquired or explored before there’s genuine clarity on which business challenge it’s meant to address. Without that foundation, there’s nothing to measure success against. - 2. No predefined success metrics
Even where a problem has been identified, the criteria for success are often left vague or defined only in technical terms. If success is measured by model accuracy rather than business impact, it becomes impossible to demonstrate value at a level that matters to leadership. - 3. No clear ownership of outcomes
AI programmes that sit between teams, partly owned by technology, partly by data, and partly by the business, rarely have a single person accountable for outcomes. Without that ownership, accountability is diffused and momentum is easily lost.
Any one of these gaps is enough to undermine a programme. All three together almost guarantee that value will remain unclear.
What do high performing FS firms do differently with AI?
The firms successfully demonstrating AI value share a common approach. They apply discipline before and during delivery: starting with a clearly defined business problem, agreeing what success looks like in business terms before anything is built, and assigning a named owner accountable for results. Many of the businesses we support have come to us after an initial wave of AI investment that felt promising but failed to produce clear returns. In most cases, the technology was sound. What was missing was the structure around it.
How do you measure AI ROI in financial services?
Measuring AI ROI requires moving beyond technical metrics and connecting programme performance to business outcomes. In practice, this means tracking measures such as:
- Revenue impact: additional income generated or losses prevented as a direct result of AI-enabled decisions
- Cost reduction: measurable decrease in manual effort, processing time, or operational overhead
- Risk reduction: reduction in audit findings, compliance incidents, or fraud losses
- Customer outcomes: changes in satisfaction scores, resolution times, or retention rates
- Speed to value: time from deployment to measurable business impact
The specific metrics will vary by use case, but the principle is consistent. If a measure cannot be connected to a business outcome, it’s not a measure of ROI.
What can financial services firms do to improve AI ROI today?
If you are uncertain whether your current AI programmes are on track to deliver measurable value, a structured review is the right starting point.
- Review your current AI initiatives and be honest about which ones have a clearly defined business problem at their core, and which do not
- Define or redefine success metrics in business terms. If your current measures are primarily technical, that needs to change
- Assign ownership. Identify who is accountable for each programme delivering business outcomes, not just for the technology performing correctly
- Assess integration. Ask whether AI is genuinely embedded in how decisions are made, or whether it exists as a separate layer the business works around
These steps won’t resolve every challenge, but they will quickly reveal where the most significant risks to value lie.
What’s next for AI in financial services?
AI investment isn’t going to slow down, but the tolerance for unproven returns is narrowing. Boards are asking harder questions, regulators are paying closer attention, and the firms that can clearly demonstrate what their AI investment has delivered will be better positioned than those that cannot. The gap between activity and value will only become more visible over time.
//CTA
At Orcan, our Data and AI work is built around the same principles we apply across all transformation delivery: senior-led, outcome-focused, and with clear ownership from start to finish.
If you are uncertain whether your current AI programmes are on track to deliver measurable value, a short discovery call is the right place to start. No preparation needed. No commitment required.