The Foundational Importance of Data Governance in Financial Services

[Asset managers and wealth management firms have a lot of data, but much of it is inconsistent, unstructured, or duplicated. It has been contended that the majority of these data issues are actually symptoms of a lack of proper governance from the very beginning of the data lifecycle and are the root of many technology challenges. Governance must be established at the point of data inception, not just as a management layer after the data has been collected. This is because, ultimately, it is governance that determines – or should determine – the process for how to create, clean, and structure the data.

Take, for example, AI agents. They already do useful work, but poor inputs (e.g., messy RFP libraries, fragmented CRM notes, inconsistent client records) limit accuracy and trust. This supports the idea that advanced modern technology like AI, in its application to our industry, is only as good as the data behind it.

To better understand the foundational importance of data governance, we reached out to JT Tripple, Enterprise Sales and Microsoft Cloud specialist at HSO – a global IT services and consulting firm with a rare combination of financial business application and data expertise. The firm was recognized for delivering transformative customer engagement solutions powered by Microsoft Cloud and AI technology by winning the “2025 Microsoft Dynamics 365 Sales & Customer Insights Partner of the Year Award” for demonstrating excellence in innovation and implementation of customer solutions. We asked JT to share their knowledge and data expertise with us.]

Hortz: Can you more fully explain the concept of data governance? What exactly is entailed and why is it of such foundational importance?

Tripple: When I think about data governance, I think of it as a combination of decision rights plus accountability over the data. It outlines how data is being defined within the firm – who owns the data and who can use it – and also how compliance is being enforced across the organization.

Data governance should also align with business outcomes. Firms should always be thinking in terms of how data impacts what the business is actually doing, not in a vacuum. Policies need to be consistent across the entire organization – the front, middle, and back offices. This is obviously critical in financial services with its regulatory and client trust requirements.

But what I think is most important about data governance is that it is foundational, because everything else sits on top of it. AI, reporting, compliance, analytics, risk management…all of these areas will fail or stall without well-defined and well-controlled data.

Data governance is outcome-based. It is not the data itself that ultimately matters; it is the outcomes that we get from it, the AI it informs, and the reporting it drives. If your data is not well-governed, all of the outcomes are put at risk and are questioned because the data underpinning them cannot be trusted.

For example, if you are developing a chatbot using AI to interact with your data, but you can’t trust the answers the chatbot is giving you, what’s the point of even investing in the tool in the first place? That’s what we are always trying to ensure – that the results of what we are doing with the data can be trusted.

Hortz: What are the most common data quality failure points? Why do they tend to persist and are often rooted in governance problems?

Tripple: To address the issue of data quality failures, there must first be agreement on how terms are defined. There are many terms that are used inconsistently within an organization, which can lead to problems. Take, for example, the word, “client”. What does that mean? Who is considered the client in each area of the business? That can mean different things to different teams, which sets up distrust between groups who believe other groups are not correctly representing the conversations happening on the ground.

Another reason for quality failure is fragmented data. People talk about how data is “siloed” – the idea being that data within the organization is spread across different business applications and functions. There are different lines of business and products or services being offered. Because of this, you can have product data sitting in one place, operational data sitting in another, and risk data in yet another, with no accountable owner driving consistent standards across it all. A significant challenge and potential failure point arises when data is not integrated and governed consistently across the firm.

Data entry itself is another point of failure. For example, if a wholesaler is not accurately entering data into the CRM, that’s a point of failure. Key data not being captured accurately has the potential to hinder the entire organization’s ability to operate at speed, at scale, or most effectively.

These data failures tend to persist and are often rooted in governance problems because manual workarounds become the norm. What I mean by that is that many firms use spreadsheets to fix problems with data coming from CRM and finance instead of addressing the issue at the point where the data is captured. For example, someone in sales ops uses their own spreadsheet to get some reporting they need, but cannot get it natively because of the way the data is structured. This approach gets the job done for the moment, but it becomes an institutionalized workaround that ultimately causes bigger problems down the line. 

Hortz: Could you walk through the first practical steps a firm should take to start establishing a foundational data governance framework?

Tripple: First and foremost, I want to emphasize this: You have to start with a business priority. You start with a business outcome, and then you go from there. You do not start with policy; you don’t start with rules. You start by asking, “What are we trying to do with our data? What are the priorities of the business?” And then you work your way into a governance model. Governance needs to be anchored in a real initiative. It could be regulatory reporting, it could be on AI, it could be getting a 360-degree view of your client. Whatever it is, whatever that priority is, you start there, so what you are doing with governing your data actually drives tangible value.

Next, determine the critical elements of your data and assign accountable owners to those elements. Keeping with the theme of business priorities, you want to focus on high-impact data – client data, product data, transaction data, or performance data. Those are different high-impact data domains. And you want to have someone in place who is truly responsible for that data. In this context, we call that stewardship – someone who has ownership of that particular data, someone who is fully accountable.

The next practical step from there is to set up an operational model. It can start with establishing a governance council, a team responsible for making decisions around data quality and determining escalation paths for issues around how the data is being organized. It is very important to involve people in mastering data, not depending solely on tooling and technology, thinking they alone are going to solve those problems. You need to have people in the cycle, making decisions and establishing guidelines, rules, and policies.

It’s challenging to govern data. It’s not fun. Nobody wants to wake up and go in and put out a bunch of policies and rules in place around data, but doing so matters when it comes to strategically reaching goals for the business and high-level engagement with clients.

Hortz: How do you advise your financial services clients to measure the ROI on improving data governance? Are there tangible metrics beyond AI project success that firms can point to?

Tripple: Asking how you can tell if your data efforts are working is the right question, but a really hard one to answer. It is hard to get stats that show demonstrable ROI on data governance. There is no benchmark to work off of. But we know this much: bad data leads to bad decisions. IBM did a study estimating that US businesses are losing over $3 trillion a year due to bad data. But a Gartner study found that 60% of firms do not even measure the effect of poor-quality data. Think about it. That’s over half of the firms out there that are not even trying to figure out the ramifications of using poor-quality data.   

There was another study that showed that data scientists spend 60% of their time cleaning data. That’s the wrong place to be spending their time. They should be analyzing data. They should be exploiting the data for the tool that it is, but they are spending the majority of their time – over half, to be specific – just trying to get the data shaped in a way that they can even begin to use it. 

But there are ways firms can measure ROI. One way is to reduce what I will refer to as operational friction. With well-governed, clean data, operational friction is reduced. There are fewer reconciliations needed between different aspects of the business. There are fewer manual adjustments needed. The number of audits that need to take place can be reduced.  

Another way is measuring ROI based on risk mitigation – fewer compliance exceptions, lower regulatory remediation costs, and so forth. One of the most valuable and interesting ways to mitigate risk is around revenue enablement. Think about how much faster you can launch a product if you have good data. How much faster you can get to market, reduce time to market from an analytics standpoint. You can improve your cross-sell and upsell with better client data.

Those are just a few of the ways you can determine how governing and managing data more effectively is driving ROI.

Hortz: What does “AI-ready” data look like? What is needed to transform data into an “AI-ready” state?

Tripple: To be AI-ready, you need your data to be consistent, and it needs to be well-defined. It cannot be siloed. It needs to have a consistent structure. It needs to be what we call “lineage traceable” – that is, you can explain where the data came from. What’s the source of this data? How did this data get transformed so that it is now consistent, and who owns the data? That’s the stewardship piece of data management.

Those are the key areas we talk about around having data in an AI-ready state. It is organized and clearly labeled. It is tagged and structured so AI can understand what it means. Data is not just being stored in a big bucket; we have actually done the hard work of cataloging the data.

And then the last thing I would say that makes data AI-ready – and the most important piece – is that it is accessible. Teams trying to develop AI use cases must be able to get to the data. If they cannot get to it, nothing else matters. You have to be able to open the door to give access to data, but in a controlled way. And that’s the key caveat; access must be controlled. But once that data is clean, well-defined, and well-governed, you can make it available to your AI teams, and they can run with it.

Hortz: With data governance in place, can you share your perspectives on building real-world AI use cases that feel concrete to asset managers and wealth management firms?

Tripple: The first and most important thing I want to communicate when we talk about AI use cases is that it is so critical to have what we refer to as a “human in the loop”. AI is incredibly powerful and has the potential to deliver powerful outcomes and execute complex tasks for you, but people must be involved. It is essential to have people play a role in most, if not all, AI use cases, critiquing the results to make sure they align with reality and with policy.

Getting into some specific use cases for asset managers, assistance with investment commentary is an area where firms could really benefit. AI can draft performance narratives using well-governed performance data, pull that data from various sources, and put together a thorough commentary narrative.

For institutional firms, RFP responses can be largely automated. Document intelligence and knowledge management can be brought to bear on existing RFP libraries, and that data can be used to shape and craft responses to future RFPs – and do it in a much more efficient way.

For both asset management and wealth management firms, meeting summarization is a huge use case. Agentic CRM can be very powerful for wholesalers. With Agentic CRM, instead of logging into a CRM platform and doing your work there – which most salespeople do not want to do – you use a chatbot, which enters the data into the CRM. You can use an app on your phone to record a meeting you had with an advisor or a client, then have the AI summarize notes, capture follow-up items, suggest next best actions, draft follow-up communications, and create opportunities and leads in the CRM system. AI can then draft personalized ongoing communications based on what’s in the CRM, based on portfolio data, what’s happening in the market, and people’s specific interests and risk tolerances. It could also identify tax loss harvesting opportunities, concentration risk, or even rebalancing portfolio opportunities using well-governed household-level data. With an agentic model like this, you can have all those tasks without ever having to go into the CRM platform. And it can really accelerate and help you scale your outreach.

Also, for wealth management firms, another concrete AI application would be around client onboarding acceleration. Consider all the forms that clients must fill out and the documents they have to provide. AI-driven document ingestion and extraction is one way to speed things up. And then KYC – Know Your Customer validation – can be streamlined through AI and the new account setup.

Hortz: It was noted that HSO has a rare combination of business application and data expertise. How does this dual perspective change the way you approach a problem for a financial services client?

Tripple: At HSO, we start with the business process. We ask business what outcome we are trying to drive to, not the data model itself. Because we look at dealing with challenges through an industry lens, we understand common issues like front office workflows and compliance challenges and constraints. We know what the revenue levers are. That’s a great start, but then we dig deeper to fully understand the client’s business, their unique challenges and requirements, before we look at designing a data architecture.

We also have the ability to bring an accelerated point of view to a firm’s data. While all financial firms are not the same, there are some consistencies between them. We can bring a perspective that is true across the industry, in addition to understanding firm-specific particular business drivers and outcomes.

Another thing we prioritize is connecting application configurations to governance. When we “stand up” a CRM platform, an ERP platform, or a data platform, everything is aligned to a common data model and an operational framework. This way, AI is consistent with the organization’s larger data model.

We have also developed separate business accelerators for asset managers and wealth management firms to bring a data model to the table that can work for them. There are two sides of the coin. On the one side, there are the bespoke business needs and requirements. The things that matter. The priorities. On the other side of the coin is the fact that all financial firms have similar needs for their data. Because we understand these needs and have worked with many firms, we are not starting from scratch with each engagement. Our accelerator process and structure expedites the design of a bespoke data model for financial firms to govern their data.

And finally, we are very good at translating between stakeholders. What I mean by this is that we help the chief information officers, the chief data officers, people on the technology side of the house, and the business leaders on the other side of the house to come together and align on outcomes. Governance is a growth enabler, not a restrictor; a partnership between the business leaders and the technology leaders drives better outcomes for the firm as a whole.

Hortz: Since many financial firms still require education on implementing advanced AI solutions, how do you engage them so that it resonates across the enterprise?

Tripple: As I said before, we operate on an industry-first basis – we are in the financial services industry, not just serving it. I think it’s very important that we are viewed in that way. We want to be a contributing member to the industry, which is why we participate in industry groups like the SME Forum where we participate as a vendor partner, contributing to discussions around developing AI use cases and sharing what we and other industry firms are doing with AI and data and how they are driving better analytics and better outcomes.

I think it would be valuable to point people to the assets we have created that demonstrate our commitment to providing education and resources to the financial industry – white papers that share our research; workshops that educate firms on AI use cases and building AI agents; self-assessments and other tools; and ongoing social media and blog posts.

This article was originally published here and is republished on Wealthtender with permission.

About the Author

Bill Hortz

Founder Institute for Innovation Development

Bill Hortz is an independent business consultant and Founder/Dean of the Institute for Innovation Development- a financial services business innovation platform and network. With over 30 years of experience in the financial services industry including expertise in sales/marketing/branding of asset management firms, as well as, creatively restructuring and developing internal/external sales and strategic account departments for 5 major financial firms, including OppenheimerFunds, Neuberger&Berman and Templeton Funds Distributors. His wide ranging experiences have led Bill to a strong belief, passion and advocation for strategic thinking, innovation creation and strategic account management as the nexus of business skills needed to address a business environment challenged by an accelerating rate of change.