Waters Wrap: For data managers, the new problems are the same as the old

It’s easy to get caught up in the excitement of technological evolution. There’s the relentless push to the cloud by banks, asset managers, exchanges and vendors. Trading firms are becoming more comfortable with open-source tools and protocols. APIs are now the preferred way to connect to sources of data and software. Low-code development has become one of the catchiest buzzwords to permeate press releases. Blockchain and other distributed ledger technologies won’t go away, despite the setbacks. New tools that provide context around a sea of data are becoming increasingly valuable. And it’s easier to provide context if systems are interoperable and if AI is used to help in these efforts.

I mean, if you work in the world of technology and aren’t excited (and worried) about the possibilities that large language models like OpenAI’s ChatGPT provide, then your blood runs cold, my friend.

But if you talk to data managers from across the spectrum of the capital markets, what keeps them awake at night is not a cool chatbot, but rather the boring foundational bits—data quality, lineage, timeliness, accuracy, and completeness. If you wanna use those innovative new tools mentioned above, you need to secure your data foundation. If you wanna avoid a nine-figure fine levied by a regulator, your data governance best be in order.

I’m going to write on the fly here. I’m not sure there will be a specific point—consider this more a manifesto for those unsung heroes that toil in data governance.

Data quality, lineage, timeliness, accuracy, and completeness—all of these things end up resulting in a more efficiently-run organization. In totality, what they create is an organization the runs on the concept of data as a product

When you get data professionals together and ply them with a couple of drinks, it’s not Amazon, Google or Microsoft they want to talk about—it’s all about satisfying regulatory obligations; talent acquisition and retention; getting budget to solve problems holistically, rather than with popsicle sticks and Band-Aids; and proving ROI for establishing a so-called “data-driven culture”. Do I have master and reference data centralized? Are employees buying market data from different places and is there overlap? Do I know where my customers’ data resides?

If you’re not in the thick of it, it can sound boring. But—and to drive it home—if you don’t get that data foundation in order, you’re likely to spend a ton of money on cutting-edge tools only to reap minimal results. And that’s when the CEO and CFO start asking the hard questions (if not the regulators).

Data quality, lineage, timeliness, accuracy, and completeness—all of these things end up resulting in a more efficiently-run organization. In totality, what they create is an organization that runs on the concept of data as a product, and moves from a mindset of defense to offense.

For example, as cloud becomes the raison d’être for banks, asset managers and vendors, that’s a good direction to charge. But AWS, GCP, Azure, Snowflake or IBM (depending on your cloud needs) won’t fix data problems. Maybe lessen some burdens? Save some expense? Mask certain issues? Sure, but a subpar internal data foundation will lead to subpar results.

The potential silver lining on the horizon, though, is that as fines rain down on end-users, as passive investing cuts into active and the need to find unique insights grows, as the amount of data increases exponentially, and as the hunt for talent only becomes more difficult, then the “data foundation” story becomes a much more accessible story for the C-suite to understand and embrace—data as a product, and commercializing and monetizing that product. $$$…cha-ching. If you can prove that you’re leaving money on the table by not having a full understanding of your internal staff and customers, and that can only be achieved with proper data governance, the purse strings can loosen.

Now, if we’re being honest, that silver lining has been on the horizon for…well…since the dawn of market data? I’m sure in the 1630s there was some Dutch dude wondering if there was a better way to improve the quality and accuracy for tulip bulb market data.

So it is that if a change is to come, it’s not going to arise from cutting-edge new technologies—it’s a people problem. In this case, revolution must come from the top down. This has been something that has been said ad nauseam through the years, and publicly every data manager will say their CEO and CFO are all about being data-driven. Have that conversation off-the-record at The White Horse Tavern (25 Bridge Street in the Financial District, of course!) and you can sit back and count the number of f-bombs dropped.

Additionally, as Max Bowie recently wrote, we’re seeing a “brain drain” when it comes to data professionals (sans quants and data scientists, of course). There’s no shortcut around having the right people in the right positions—people with a deep understanding of both the business and workflows; people who can define the right data model and deliver the right data products. Maybe you can outsource that, but it’s been tried many times and has failed many times.

Furthermore, you can have data stewards internally, but are they being given the time and resources to not be snowed under? Are they really stewarding data, or being asked to serve as business analysts for data migrations and new data projects? You need subject matter experts that have an appropriate level of bandwidth to think holistically, rather than jumping from “urgent” project to “urgent” project. You need to bake these people into the organization’s data products’ lifecycle.

Finally, you need to teach people. Everyone at the organization needs to have data literacy and fluency. For that, you need programs in place that are mind-numbingly boring, but that actively show the benefits to an individual’s daily routine (and salary). That costs money—again, do your CEO and CFO really care about this stuff, or are they just paying lip service?

Cloud, open-source, APIs, machine learning, natural language processing and low-code development are great (and we haven’t even discussed quantum computing/encryption), but it’s the so-called boring stuff that makes the exciting stuff work properly.

If you have good ideas as to how best to address these issues, trust me, we want to hear from you. Shoot me an email and we can chat…I can even take you out for a few drinks, should you like: anthony.malakian@infopro-digital.com.

The image accompanying this column is “Olive Trees” by Vincent van Gogh, courtesy of The Met’s open-access program.