The next is a sponsored submit from Ted O’Connor, SVP and Head of Enterprise Improvement—Promote Facet, with world fintech firm Arcesium.
Arcesium delivers a sophisticated information, operations, and analytics platform utilized by among the world’s most subtle monetary establishments, together with hedge funds, banks, institutional asset managers, and personal fairness corporations.
Each financial institution is in a special stage of its information journey. Just lately, whereas attending the InvestOps Europe convention in Paris, one of many presenters talked about that in the case of gauging the extent of confidence banking management has within the integrity of its information, 95% confidence of their information is the barometer to which they should adhere. Ninety-five % has at all times been a fascinating grade to get on a paper or in a category, however is it adequate when speaking a couple of multinational financial institution working in dozens of jurisdictions?
Just like the air we breathe, information is odorless, colorless, silent, and laborious to measure. That’s, till information is introduced subsequent to greenback indicators on a disclosure report, steadiness sheet, or interminable spreadsheet; then it turns into actual. The previous few years have seen monetary establishments grappling with immediately ballooning volumes of monetary information, not a simple ask for legacy information methods and banks that may run on scores of various methods.
The 95% confidence fallacy
Whereas a 95% confidence interval[i] in information is the goal, banks actually have solely 80-90% confidence of their information at this time. In a 2024 examine of sell-side reference information operations, over 90% reported that poor information high quality precipitated points in clearing and settlement, threat administration, and regulatory reporting, with 80% citing challenges in automated buying and selling and market connectivity emanating from inaccurate information.[ii] Furthermore, that 80-90% is a little bit of an phantasm. Right here’s the fact. Say, I’m a financial institution CTO or chief information scientist, and I’ve 80% confidence within the information that’s coming to me through any sort of transaction. I then push that information into the clearing or matching course of. Then, I push it into the settlement course of—and there’s money motion that goes together with this. That information retains getting pushed from one course of to the subsequent, to the subsequent, and the subsequent, which implies there’s slightly little bit of degeneration that occurs throughout. By the point I get to the top of my processes, I’ve 50% confidence in my information, and that little anomaly from the primary course of turns into a severe information drawback 10 steps later. Nevertheless, that is an inscrutable drawback to acknowledge, a lot much less resolve. It depends upon the robustness of the establishment’s current information and operational infrastructure, the stage of its information transformation journey, and the asset courses and constructions concerned.
In the meantime, the chance of getting it improper is excessive. On the undesirable finish of the 95% spectrum, Citi shelled out a couple of billion {dollars} in fines within the final 5 years for irregularities in its regulatory reporting information and governance failures, and responded by spending thousands and thousands modernizing its know-how.[iii] Deutsche Financial institution, Wells Fargo, and Mitsubishi Financial institution are examples of establishments which have labored by means of confidential supervisory findings referred to as Issues Requiring Consideration (MRAs) and Issues Requiring Quick Consideration (MRIAs). Many of those have been rooted in information processes. On this context, even 95% (and even when it had been a real 95%) isn’t sufficient for world banks—UBS, as an example, has a steadiness sheet bigger than the Swiss economic system. A Swiss bailout of such a financial institution is difficult. The chance must be near-zero, which implies confidence must be near-perfect.
Is AI the important thing?
AI has lit a fireplace within the bellies of buy-side and sell-side establishments alike, as they know their information home have to be to ensure that the AI home to be so as. In keeping with Deloitte, “Banks’ AI readiness is usually slowed by the information foundations that fashions rely upon. Poor infrastructure can lead to information sprawl, vulnerability, and restricted data-led innovation, limiting mannequin efficacy.”[iv] However as soon as a financial institution has their AI recreation in place, it might play a pivotal function in bringing order to the information chaos. There are a number of information high quality administration features that AI brokers are already serving to with. For instance, one monetary establishment not too long ago leveraged generative AI to automate information lineage seize and metadata technology, reaching 40% to 70% productiveness positive aspects in particular duties.[v]
AI presents ready-assistance for unstructured information, specifically. If managing structured information is like sorting pre-labeled packages, managing unstructured information with AI is like immediately studying 1000’s of handwritten letters, figuring out key info in each, and organizing these info right into a searchable spreadsheet—a process not possible for people at scale. However, once more, the artwork of the attainable in the case of AI will come again to information high quality; it would require establishments to centralize their information administration capabilities, with an emphasis on instruments that help sturdy information lineage and reporting accuracy.
The 100% information confidence paradigm
Having a 95% information confidence barometer presents a number of pitfalls when executing tech transformations. Regulatory concerns, information governance challenges (particularly with unstructured information), surging market volumes, personal credit score, and the adoption of AI within the monetary companies {industry} are forces that can’t be ignored. Realistically, banking leaders have to preserve their eyes on the 100% prize for high quality information administration.[vi] All people beneath the roof will do a greater job in the event that they belief that the knowledge they do their jobs with is dependable, well timed, and exact.
[i] Investopedia, Could 6, 2025. https://www.investopedia.com/phrases/c/confidenceinterval.asp#toc-explain-like-im-five
[ii] Acuity Information Companions, November 2024. https://belongings.ctfassets.web/cy2jgjrgaerj/5V6yrRfzYZU1LXqUgvulAD/ed8d59627717a3fafe96f36123d36e8e/increasing-efficiency-in-sell-side-reference-data-management-fow.pdf
[iii] Banking Dive, July 11, 2024. https://www.bankingdive.com/information/citi-occ-fed-135-million-penalties-2020-orders-data-quality-risk-management-control-fraser-hsu/721061/
[iv] Deloitte, October 30, 2025. https://www.deloitte.com/us/en/insights/{industry}/financial-services/financial-services-industry-outlooks/banking-industry-outlook.html
[v] BCG, Could 6, 2025. https://www.bcg.com/publications/2025/tech-banking-transformation-starts-with-smarter-tech-investment
[vi] Arcesium, February 2, 2026. https://www.arcesium.com/assets/driving-trusted-data-framework-for-banks?utm_source=one-off&utm_medium=show&utm_campaign=MC-2026-Q1_SS-Knowledge-High quality-To-Do-Record&utm_content=finovate-sponsored-article
Views: 93
