Analytics as asset: why financial services leaders must master data-driven strategy

The race to embrace AI is showing no signs of slowing. Financial services firms’ investment in AI is expected to almost triple to $97bn in 2027 from $35bn in 2023, according to a World Economic Forum report. Yet most AI initiatives so far have remained trapped in experimentation mode. According to Deloitte, 72% of firms say fewer than 30% of GenAI experiments reach production, meaning most stall at the experimentation stage.

Struggling to scale

There are several reasons why financial services firms are struggling when it comes to moving beyond experimentation and scaling up their AI projects. According to Deloitte, three of the top four barriers financial services firms face when deploying GenAI tools are regulatory compliance, difficulty managing risks and the lack of a governance model.

“When it comes to regulatory compliance, firms want to be certain there is transparency around how AI models work so the outputs are verifiable and that any customer data is secure,” says Jawwad Rasheed, global transformation lead at Alteryx. “Regulators also expect firms to put measures in place to improve comfort around how these AI tools are used – such as access controls and ensuring there is always a human in the loop.A lot of financial services firms are hesitant in increasing the applications of AI for external disclosures unless the results can be explained.”

In turn, there is a greater appetite to adopt AI tools for internal operations, such as interrogating data to help identify anomalies or to spot trends, which poses less financial and reputational risk to organisations.

From a customer-facing perspective, chatbot-type solutions are also making more headway, as brands attempt to respond to customer queries or route requests in a seamless and more efficient manner. Similarly, the use of AI to streamline customer operations has seen huge advancements, such as simplifying loan origination and client onboarding.

“It comes down to a clear risk assessment around AI adoption. Teams are dipping their toe in the water and experimenting but there is still hesitancy to put such projects into production until they are comfortable that there are internal procedures, policies and resources available to manage business risks and address issues when things break down,” says Rasheed.

To that end, just 25% of financial services firms say their organisation is ‘highly’ or ‘very highly’ prepared when it comes to AI risk and governance. This matters because without adequate data governance controls in place to manage data quality, accessibility, security, data ownership and standards, a firm’s AI projects may be vulnerable to data inaccuracies or bias, inviting regulatory pushback.

“It’s the lack of preparation across many of those things that has led people to say we’re probably not ready to push this into more of a live environment and to be comfortable with the results that we get,” says Rasheed. “We’re seeing that building strong foundations is critical to effectively advance the AI journey – and this comes back to strengthening core principles around data governance and management, data literacy augmentation and data infrastructure investments.”

Experimenting with purpose

Another reason why firms are often still stuck in the experimentation phase is that many are unsure how to experiment with purpose.

“People often don’t know how to experiment,” says Rasheed. “If you’re experimenting with building a car, you should know what a car looks like or what you’re working towards. When you don’t know what’s available to you or what you could achieve, then it’s difficult just to tinker.”

Firms therefore need to have better direction around what the possibilities are and what tools are available, and then understand how to join those up in a way that is most relevant for their teams.

The problem for firms that see their AI projects stall at the experimentation stage is that it creates an innovation bottleneck, which can impact in two key ways.

“Firstly, it’s a cost just to continue to spend time and resources to experiment but not put into practice and deliver returns in your day-to-day business,” says Rasheed. “The second is frustration. People are unsure how to use AI tools or how they could benefit, and they get frustrated and move away.”

This can then lead to fragmentation where people start using unauthorised tools or find their own use cases without the right controls in place, resulting in AI sprawl and pockets of activity that are not connected.

From compliance burden to competitive edge

Stalled AI projects don’t just create inefficiencies, they can also put firms at a competitive disadvantage that can ultimately drag on growth.

“If you’re not using the tools and technologies to remain competitive, other organisations will take the leap and step ahead,” says Rasheed. “People who understand how AI can best be used will become hugely valuable, and those organisations will thrive.”

Against this backdrop, many firms are focused on AI adoption that not only focuses on reducing operational costs, increasing agility and improving price-performance of services, but also drives better business decisioning and top line returns.

“Take FP&A, for example. Correct application of machine learning models can drive smarter and more accurate forecasting. If this is complemented with LLM integrations that can better explain the numbers and variances in natural language, then businesses can make quicker and more informed decisions to drive returns,” explains Rasheed.

“That’s not to say that streamlining existing operations isn’t a priority, it’s more about recognising that AI adoption can stretch to improve a wider spectrum of metrics.”

Another area where LLM integration can give firms a competitive advantage is tax compliance. Use of an LLM can be extremely advantageous to sift through complex and voluminous policy documents to assess adherence to compliance requirements and internal standards. This is a time-consuming process that would otherwise involve compliance teams having to manually search through hundreds of documents and PDFs.

“The use of LLMs just accelerates how teams work and takes away all of the burden of trying to improve the level of compliance,” says Rasheed. “This is the kind of uptake we’ve seen around AI use, where it’s a complementary component of what people do today.”

The upside from this approach is clear. A Citi report on AI in Finance last year predicted that AI could boost banking industry profits by $170bn by 2028 through productivity gains driven by automating and streamlining operations. Banks can optimise up to 66% of time spent on operational, documentation and compliance-related activities by using AI tools, according to Capgemini.

This optimisation can help free up employees to focus on more important work. For example, analysts can save 8.6 hours a week on average by using AI and automation tools, according to Alteryx’s State of Data Analysts in the Age of AI report.

It is not just time-saving benefits either. By using AI tools, firms can identify missed business opportunities, such as cross-selling customers different products.

Laying the foundations for sustainable AI adoption

However, the real advantage comes when firms use AI to fundamentally change the way they work and to reshape their organisations.

Rasheed says Alteryx is helping financial services firms do this by providing a unified data platform that simplifies data analytics transformation, enabling firms to become more data literate without going through decades of coding and development challenges.

“We are helping people take steps in leaps and bounds to not only be more comfortable with data analytics, but also help orchestrate workflows with LLM integrations,” Rasheed says.

Those LLM integrations mean firms can use Alteryx as a central data clearing house for information that goes into and out of an AI model to make sure it is secure and compliant and has a human in the loop.

By using a platform like Alteryx, therefore, financial services firms can embed transparency, automation and auditability into their AI workflows, enabling faster innovation, improving regulatory compliance and ultimately driving more sustainable growth.

Financial services firms are facing a twin dilemma. Despite spending ever-greater sums on compliance, fines for regulatory failings are increasing. According to Fenergo, firms were on the wrong end of $4.6bn in regulatory penalties in 2024, a staggering 522% increase compared to the previous year. At the same time, firms are spending $206bn globally on financial crime compliance alone, according to Forrester. This paradox of higher compliance costs but rising fines is shining a spotlight on outdated compliance practices where firms continue to rely on manual, siloed processes.

This matters because regulatory demands are not slowing, with increased expectations on firms around anti-money laundering, sanctions and transaction monitoring. This is putting even more pressure on stretched compliance teams, with those traditional processes creating compliance gaps and exposing firms to heightened regulatory risk.

“It’s increasingly difficult to keep on top of all the various rules and regulations without investing heavily,” says Oli Greaves, head of risk and compliance at digital car finance lender Carmoola. “There’s more of everything now: more rules, more data and more enforcement. This means that monitoring now needs to be continuous and proactive.”

Consumer Duty, for example, ramps up expectations on firms to proactively monitor the outcomes all customers receive, Greaves says. The problem they face is that many firms rely on antiquated, disconnected systems with too many processes that are dependent on email and spreadsheets, he adds.

“Alert volumes outpace headcount and talent is tight, especially where compliance meets data,” Greaves says. “That makes timely, trusted MI hard to produce. Horizon scanning is also a significant challenge. Many companies are now using AI to scan regulatory documents to find relevant information for their business.”

This underscores one of the biggest challenges that compliance teams face. Even if they had the budget to hire more staff, there are not sufficient numbers of qualified individuals to keep up with expanded compliance workloads.

“Although the talent pipeline into compliance has been increasing in recent years, there continues to be an under-supply of experienced individuals who understand the fundamentals of compliance and don’t operate with a tick-box mentality,” says Rachel Aldridge, managing director for regulatory and compliance solutions UK at IQ-EQ, an investor services business. “While the FCA is pushing back on what it sees as under-qualified individuals being appointed to key SMF 16 and SMF 17 roles, the pool of suitable individuals is small.”

New recruits may also opt to join firms that are investing in modern data and compliance platforms, exacerbating the problem for firms that are still wedded to manual processes.

“It makes jobs harder and, at a minimum, leads to frustration,” says Aldridge. “Firms with these issues will struggle to attract and retain talent.”

Increasing headcount and modernising compliance systems is only part of the challenge firms face in a world of rising compliance costs and steeper regulatory fines. Firms also need to ensure they have a proper compliance culture embedded into the way they work.

“The biggest challenge for financial firms isn’t just implementing compliance technology or frameworks, rather it’s building trust in the system as a whole,” says Simon Thompson, head of AI, ML and data science at GFT, a digital transformation consultancy. “Compliance obligations aren’t met by infrastructure alone. They depend on organisational culture, discipline and capabilities to make these systems effective.”

The consequences of getting this wrong are significant. Not only do financial services firms face the risk of fines, compliance failures can also result in severe reputational damage and shareholder backlash. New regulations such as Basel III Endgame, Consumer Duty and ESG reporting requirements are also adding to compliance workloads and increasing regulatory risk. For firms that are still relying on manual workflows and legacy systems, this is only going to get harder for them to manage.

“Sustaining compliance performance through ad-hoc, manual or siloed processes is a big challenge,” says Thompson. “These systems often rely heavily on a few key individuals. A single individual leaving or falling ill can compromise the entire compliance workflow, if they’ve been the glue that’s held it together for years. Many of these manual processes are also off-shored and/or outsourced, which amplifies the fragility of the overall system.”

These manual processes, such as spreadsheet reconciliations or one-off reports, coupled with fragmented data systems, increases the chance of errors and regulatory mishaps, and is the main culprit for the paradox of rising compliance spend and financial penalties.

To counter this, firms need to start automating reporting workflows to ensure processes are fully auditable at every step in the chain. This can help firms manage increased compliance volumes without needing to constantly hire more staff, while also reducing the risk of fines for reporting failures.

“Automation gives you repeatability, timestamps and lineage, which is exactly what auditors and supervisors want to see,” says Greaves. “It lowers error rates, speeds up reporting and frees skilled people to focus on tricky cases or vulnerable customers, not cutting and pasting CSVs. Start with the high-impact returns and outcomes MI, then layer on targeted quality checks. The result is cheaper compliance and better customer outcomes.”

Automating routine work is not just about cost savings, it also makes it easier for firms to scale workflows more consistently.

“Automated systems also enforce standard processes, which reduces the risk of variation in how different teams, in different locations may handle the same task,” says Thompson. “This allows firms to respond quickly to new regulatory requirements, at a quicker pace, which in turn reduces their operational risk.”

By investing in a platform that enables financial services firms to automate reporting and standardise workflows in this way, firms can break the cost/fine paradox by cutting compliance spend and reducing their risk of being on the wrong end of a regulatory penalty.

The persistent use of manual tools such as spreadsheets and other end-user computing (EUC) systems is increasing the risk of compliance failures for financial services organisations, exposing them to rising regulatory fines and potential reputational damage. In the first eight months of the year, the Financial Conduct Authority (FCA) issued more than £43m in fines for reporting failures. With regulatory scrutiny increasing around data governance, EUCs remain a hidden weak spot.

To start with, EUC tools like spreadsheets are difficult to monitor and are more prone to user mistakes, such as mis-keying data or different users working on the wrong versions.

“Spreadsheets are great in a pinch, but they’re fragile,” says Oli Greaves, head of risk and compliance at digital car finance lender Carmoola. “You get version creep, hidden formulas and light controls, so small slips can turn into bad decisions and shaky regulatory returns. The risk of human error is more pronounced when manual spreadsheets are used, which can result in data inaccuracies and breaches that can be hard to audit and unpick.”

This creates the potential for compliance black holes where audit gaps increase regulatory risk for firms and the likelihood of non-compliance penalties.

“Under MiFID II and Basel III requirements, firms must demonstrate complete data lineage and calculation transparency,” says Dean Clark, CTO at GFT, a digital transformation consultancy. “However, in many cases, regulatory capital calculations are performed through chains of spreadsheets across multiple departments, making it incredibly difficult for auditors to be shown precisely how a specific calculation was reached upon and who was involved.”

For regulators, if something is not recorded and therefore can’t be evidenced, in their eyes it didn’t happen, says Rachel Aldridge, managing director for regulatory and compliance solutions UK at IQ-EQ, an investor services business.

“This black-or-white approach to regulatory oversight means that using manual processes and local tools creates risk,” she says. “For example, if a firm cannot systematically show that consumers are at the heart of its decision-making, then it may fail regulatory scrutiny of its Consumer Duty implementation.”

Another risk from manual EUC usage is ‘key person dependency’, when entire regulatory reporting processes rely on one analyst’s undocumented spreadsheet macros, says Clark.

“If that individual leaves, firms face potential reporting failures and regulatory fines,” he says.

Despite these risks, EUC tool usage persists. According to Alteryx’s State of Data Analysts in the Age of AI report, 76% of analysts surveyed still rely on spreadsheets for data preparation tasks. Not only does that increase the chance of errors, it is also time-consuming: 45% of data professionals surveyed still spend more than six hours a week on data cleansing and preparation tasks.

“It can sometimes be difficult to invest in the longer term where you have a solution that works at the moment,” says Greaves. “Bringing in new tools may make things faster and more efficient in the long run, but can be painful in the short term to set up, test and train teams.”

Cost and cultural challenges can also be limiting factors for firms to move away from traditional EUC tools and implement a more comprehensive compliance system.

“In a fast-moving environment like financial services, there are always competing priorities,” says Aldridge. “As the system would still need to be operated by humans with judgement, it’s often difficult to prove that these systems are able to save cost and/or headcount, and consequently they may be deprioritised.”

With regulators demanding that firms make EUC control part of their operational resilience plans, non-compliance can have major consequences, as the millions in regulatory fines handed out this year underscore.

“There have been big, well-known issues that have come from simple spreadsheet problems, like copy-and-paste errors or file limits being hit,” says Greaves. “For example, in 2012, JPMorgan’s ‘London Whale’ losses were made worse by error-prone, manual modelling. One slip in an EUC that feeds regulatory returns or customer MI can snowball into complaints, fines and headlines.”

More recently, Barclays faced a £26m fine in 2022 for transaction reporting errors, many of which stemmed from manual data processes. But it was not just regulatory penalties the bank faced, it also resulted in client confidence eroding and increased scrutiny from institutional investors questioning operational controls, says Clark.

This is going to become even more important for firms to address as a wave of new regulations come online this year that will increase reporting requirements, including Basel III Endgame rules, Consumer Duty and expanded climate and ESG disclosure requirements.

Financial firms are therefore under more pressure than ever to reduce their EUC usage by placing stricter controls on when EUC tools can be used and investing in a platform that can help unify data workflows and eliminate regulatory blind spots.

“Companies should treat EUCs like mini systems and set some guardrails, such as standardising controls and moving core logic and data away from spreadsheets onto controlled platforms that support audit and lineage,” says Greaves. “Do this and you cut errors, speed up reporting and improve outcomes.”

Ultimately, as regulatory compliance becomes ever more dependent on connected data systems, EUC tools will become increasingly unsustainable for firms to use.

“Modern financial services require automated, auditable and scalable data platforms that support efficient operations,” says Clark. “The era of ‘spreadsheet governance’ is a convenience the industry cannot afford and should not tolerate.”

For Clark, this is not just a risk mitigation exercise, it can also help give firms a competitive advantage by enabling faster innovation cycles and more agile responses to regulatory and market demands.

Therefore by moving to a platform that supports automated reporting workflows and end-to-end audit trails, financial services firms can ensure they can keep pace with ever-growing data governance and compliance obligations, while also improving commercial outcomes for their business.