The rhetoric around data as a strategic asset is now well established across capital markets. But as AI adoption accelerates and pressure mounts to deliver tangible outcomes, the real challenge is turning that rhetoric into operational reality.
Access to data is no longer the constraint. What matters now is how effectively firms can organise, govern, and deploy it in ways that deliver measurable business value, which requires more than pipelines and platforms. It demands a shift in mindset, where data is not just accumulated or provisioned, but treated as a product: discoverable, trusted, and fit for purpose.
This evolution touches every part of the data lifecycle. From metadata and entitlements to lineage, accessibility, and quality controls, firms are being forced to rethink how they structure, manage, and expose their data assets. At the same time, organisational models are shifting, combining centralised enablement with decentralised ownership to support faster iteration, scalable AI, and platform-wide consistency.
So how are firms moving from intent to execution? What are the architectural, operational, and cultural changes required to make data a true differentiator? And what does it take to build the foundations for trustworthy, scalable intelligence in an increasingly competitive and data-driven market?
Owning the Strategy, Not the Stack
To differentiate through data, firms are rethinking not just what technologies they adopt, but how they assemble them. The traditional build-versus-buy model has given way to an approach where firms retain control over their data strategy while leveraging modular components to stay agile, scalable, and responsive.
“Even five or ten years ago, it was a binary decision: build or buy,” notes Zack Helgeson, Head of Product at Canoe Intelligence, the alternative investment data solutions provider. “Now, it’s about composing your architecture from modular components. Firms don’t necessarily want to own the data but they do want to own their data strategy and make sure they can scale without being locked into a rigid vendor stack.”
This architectural shift supports a broader transformation: from managing data as an asset to managing it as a product. In practice, that means ensuring datasets are clearly described, quality-checked, and discoverable by those who need them. It also means investing in metadata infrastructure and exposing data through well-documented APIs to support integration, experimentation, and automation across business units.
Seamless integration is now a critical success factor. “Whether it’s via API, SFTP, or direct connectors, the priority is making the data accessible where and when it’s needed,” says Helgeson. “The velocity of data – getting it from ingestion to application quickly – is becoming a real differentiator, and seamless integration plays a big part in that.”
The emphasis is no longer on raw ownership, but on how quickly and effectively firms can put data to work. “What differentiates firms today isn’t owning the data, it’s what they can do with it,” says Paul Humphrey, CEO of BMLL, the independent provider of historical market data and analytics. “Engineers and quants don’t want to spend time cleaning up messy inputs; they want to build value on top. That’s the new differentiator.”
This shift in mindset also demands a platform foundation that supports both speed and trust. “It’s about accelerating time-to-value and enabling firms to experiment and iterate faster,” Humphrey continues. “Fail fast, refine, deploy. But it’s also about consistency. Regardless of whether you’re doing algo development, TCA, or something else entirely, what firms are asking for is the same: accuracy, quality, timeliness, and adherence to market standards. Once those foundations are in place, they can focus on differentiation at the strategy level.”
Cloud-native, API-first platforms are making this possible by decoupling access from ownership and allowing teams to innovate without being tethered to legacy infrastructure. But true consistency remains a challenge. “Many firms claim to be API-first,” says Gus Sekhon, Head of Product at FINBOURNE Technology, the data management solutions provider. “But building a truly API-first architecture means the user interfaces must be powered by the same underlying methods that are exposed internally. That consistency is essential.”
Ultimately, owning the stack is no longer the goal. Owning the strategy and ensuring the architecture supports it, is what enables firms to translate data into competitive advantage.
Fit-for-Purpose Data Quality
As firms seek to operationalise AI and analytics at scale, data quality remains a central concern, although the definition of quality is evolving. Rather than pursuing an abstract ideal, many institutions are adopting a more pragmatic, risk-based approach: assessing data based on whether it’s fit for purpose within specific contexts.
This shift reflects both operational necessity and strategic maturity. Perfect data is often unattainable, and efforts to achieve it can delay progress. Instead, firms are segmenting use cases by criticality. For high-impact applications such as pricing, regulatory reporting, or model training, rigorous quality controls and lineage tracking remain non-negotiable. But for exploratory analytics or internal productivity tools, indicative data can often deliver sufficient value to move initiatives forward, provided it is clearly labelled and contextualised.
“The cost of having brilliant engineers spend time cleaning poor quality data is enormous,” says Humphrey. “It’s like hiring Picasso to paint your living room. The real cost of market data isn’t just in purchasing it, it’s in maintaining, normalising, reconciling it. You don’t want to lie awake at night worrying about symbology mismatches. You want confidence that the data is curated, correct, and ready to use.”
This approach allows firms to prioritise investment more intelligently. By identifying where poor-quality data poses genuine business or compliance risk, teams can focus resources where they matter most. It also enables incremental delivery: launching lower-risk use cases with existing datasets, learning from those experiences, and using that momentum to build the case for longer-term data remediation or platform enhancements.
Helgeson agrees that quality and governance should be viewed as foundational. “It’s easy to view governance as a compliance burden, but the firms we see succeeding treat it as a strategic asset. Having consistent processes for managing disparate data sources, applying metadata, and enforcing quality controls is what enables firms to experiment – connecting data sets, layering in analytics, testing hypotheses – without creating technical debt.”
Ultimately, quality is not a fixed state but a function of context. Firms that embrace this view are better positioned to move faster, allocate resources more effectively, and deliver outcomes with greater confidence, without stalling innovation or over-engineering solutions.
Embedding Control and Governance
As AI adoption accelerates across capital markets, data governance is under renewed scrutiny, not just for regulatory compliance, but as a foundation for trust and agility. Traditional governance models, often centralised and enforcement-heavy, are struggling to keep pace with the demands of AI development, where speed, iteration, and domain knowledge are critical to success.
To address this, firms are rethinking both the structure and language of governance. A growing number are shifting responsibility to the first line – those closest to the data – while central functions focus on tooling, standards, and oversight. This decentralised model preserves control while enabling faster, more responsive decision-making.
“There’s a tension between enabling non-technical users and ensuring governance,” says Sekhon. “That’s why we embed fine-grained entitlements in our platform. Access can be restricted at the API level, at the data domain level, at the storage level, whatever granularity is needed. We’ve found that clients want to democratise access but still retain control over who can operate on the data and in what way.”
Governance frameworks must also be accessible to non-specialists. Language and usability matter. Abstract terminology and overly technical controls can alienate users and drive workarounds. By contrast, clear guidance, intuitive workflows, and embedded tools help ensure that policies are followed without creating unnecessary friction.
Lineage tracking, usage monitoring, and entitlements remain essential, but they are no longer enough. Governance must now be embedded into daily workflows, not enforced from the outside. At its best, governance enables rather than restricts. As Sekhon puts it, “Firms are under pressure to differentiate. They can’t just replicate what everyone else is doing. They need to offer something distinct. And that requires a data architecture that supports agility, experimentation, and scale. Enabling that kind of responsiveness is where data becomes a true differentiator.”
Engineering for Intelligence
Building intelligent systems, whether for alpha generation, automation, or operational decision-making, depends on robust, responsive data engineering. And increasingly, that means solving for trust, speed, and adaptability in tandem.
“We’re seeing a strong push for consistency across environments, especially between real-time systems and cloud-based research or backtesting infrastructure,” says David Taylor, CEO of market data solutions vendor Exegy. “If the production environment and the research environment are misaligned, it introduces variance and undermines confidence in the results. Bringing historical and real-time environments together, so you can go from a backtest to a deployed strategy with minimal reengineering, is key to agility.”
In parallel, evolving cloud strategies are reshaping how firms handle and interact with operational data. “Over the past 12 to 18 months, we’ve seen many firms begin to make more tangible progress toward the kind of long-term data strategies they’ve been discussing for years,” says Tim Bosco, Managing Director for Technology and Data Services at BBH, the global financial services firm. “One major driver is the growing pressure to invest in AI and machine learning capabilities. These tools are powerful but extremely data-hungry, and their effectiveness hinges on access to high-quality, well-structured information. In parallel, we’re seeing a shift in how firms think about cloud adoption. It’s no longer just about hosting CRM systems or peripheral applications, firms are now actively moving core operational data into cloud environments such as Snowflake or Azure.”
As these architectures evolve, firms are exploring ways to interact with data in place rather than moving it repeatedly across platforms. “Regarding cloud platforms, firms are now looking at how they can move from pulling data into their environment to interacting with it in place, where it lives,” explains Bosco. “That’s highly efficient in theory. But the practical challenge is that clients still need to contract and integrate with multiple cloud providers, each with their own protocols. That introduces complexity and cost.”
The challenge of unstructured and semi-structured data also persists. “Our approach has been to focus on building workflows and visualisation tools that allow subject matter experts to apply their knowledge at the right stage,” says Bosco. “There are providers using AI and machine learning to assist with this, while others are taking more service-led approaches. It remains a major opportunity – and pain point – for asset managers.”
Engineering for AI also brings new operational demands. Models must be trained, validated, deployed, and monitored, often across different environments and regulatory regimes. This requires consistent lineage, clear versioning, and real-time observability. It also requires that the data fuelling these models is not only high-quality but transparently managed.
“Time to market – how quickly firms can move from idea to execution – is becoming a key differentiator in itself,” notes Taylor. “Even small inconsistencies like symbol mapping can derail the process. So solving for those practical data engineering issues is critical. It’s not glamorous, but it directly impacts how quickly you can innovate and respond to opportunities.”
Meanwhile, firms are refining how they assess the business impact of their data investments. “On the revenue side of ROI evaluation, the conversation differs between sell side and buy side,” says Taylor. “For sell-side clients, it’s often about execution quality. They measure that by price and market impact, specifically, how many immediate-or-cancel (IOC) orders get filled in the first wave. That’s a proxy for latency, connectivity, and targeting precision. The buy side, by contrast, is focused on returns. Did the data help them discover durable alpha? Can they validate that over multiple market cycles?”
For firms that get the engineering right, the payoff is not just speed or scale, it’s confidence. It’s the ability to act on insights without hesitation and to deliver trusted outcomes in an environment that demands nothing less.
Trustworthy Data, Trusted Outcomes
In the race to adopt AI and analytics at scale, capital markets firms are rediscovering a foundational truth: data isn’t just the prerequisite, it’s increasingly the differentiator. But turning that potential into performance requires more than aspiration. It demands execution.
From composable architecture and fit-for-purpose quality to embedded governance and robust engineering, the firms pulling ahead are those treating data not simply as infrastructure, but as product. They are aligning platforms with organisational design, balancing central enablement with domain ownership, and building the operational muscle to deliver trusted outcomes at speed.
“Unless you get your data architecture right, you can’t derive value from the tools you layer on top,” says Sekhon. “Whether it’s AI, business process automation, analytics, or visualisation, none of it works unless the underlying data is high-quality and accessible. And it’s not just about the data quality itself. It’s about the quality of the services that deliver and expose that data.”
Looking ahead, firms that can engineer trust into their data pipelines, consistently, transparently, and at scale, will be best positioned to turn insight into action, and action into sustained advantage.
Subscribe to our newsletter
