Transforming financial decisions through data analytics
Read time: 6 minutes
Executive Insights
- Lending has always been a disciplined process: move information through a workflow until a decision is made. What has changed is not the process — but the behavior of data. Where it lives. How quickly it moves. Whether it is trusted. And most importantly, whether it shows up inside the workflow when decisions are made.
- Banks are not short on data. They are short on usable, trusted, workflow-ready data. Research consistently points to the same constraint: fragmented systems prevent a unified view and slow execution. [1] Even when institutions invest in data initiatives, value creation lags. Many organizations still struggle with data quality, integration, and governance — core issues that directly impact lending outcomes. [1] [kpmg.com]
- At the same time, lenders are actively pursuing new signals because traditional credit files do not tell the whole story anymore. LexisNexis reports 67% of lenders believe access to more alternative data would help them approve more worthy borrowers, yet only roughly half have a plan on how to use that alternative data. [2] [lexisnexis.com]
- Finally, agentic AI is shifting from concept to operating model. McKinsey reports that when banks rewire a frontline domain end to end with agentic AI, they see 3% to 15% higher revenue per relationship manager and 20% to 40% lower cost to serve. [3] The point is not “replace people.” The point is to stop wasting expert time moving and rechecking data that the institution already owns. [mckinsey.com]
The takeaway is clear:
The competitive advantage is no longer access to data. It is the ability to operationalize it — consistently, at scale, inside the lending workflow.
The Evolution of Lending: What Actually Changed
In banking, we run processes. Those processes are, at their core, controlled data movement. Historically, we inserted humans with specialized skills to keep the data moving across the lending lifecycle. The legacy pattern still looks familiar: collect, re-enter, validate, hand off, repeat. Humans did not only underwrite. Humans also stitched together systems that were never designed to cooperate.
…and we evolved…
We digitized steps and added tools, but many institutions kept the same operating model: origination, underwriting, decisioning, documentation, then monitoring. Manual handoffs and re-entry still show up more than they should, especially when data lives in multiple systems and teams rely on workarounds.
…or did we?!
If your credit team spends a meaningful chunk of the day chasing documents, reconciling versions, and rechecking facts that were verified last year, you did not modernize lending. You digitized friction.
McKinsey describes the same issue on the frontline: admin load and wasted motion that keep bankers away from customers. They report that, in many commercial banks, relationship managers spend only 25% to 30% of their time in client dialogue. [3] That is not a talent problem. It is a workflow and data problem. [mckinsey.com]
Why This Conversation Matters Right Now
Borrower expectations: Borrowers expect faster decisions and clearer communication, shaped by digital experiences outside banking.
Risk and scrutiny: Credit teams face increased pressure for consistency, transparency, and auditability.
Fragmentation: Data fragmentation forces rework and slows decisions. KPMG explicitly escribes fragmented, siloed systems as a common blocker. [1]
Modernization gap: More than 80% of banking respondents in recent surveys cite privacy and risk, data quality, and legacy integration as top concerns, which is exactly where transformation efforts stall. [1] [deloitte.com] [kpmg.com]
From Data Visibility to Data Execution
The Problem: Static Snapshots
Traditional lending runs on snapshots: capture borrower data once, validate it once, and treat it like truth for the life of the loan. Modern lending requires connected, contextualized, continuously refreshed data that supports decisions inside the workflow.
This is where many banks stall. They centralize views, but they do not change the work. KPMG’s data supports that gap: only 35% of respondents across industries say they have created value from data products so far. [1] Visibility without execution is not transformation. [kpmg.com] [kpmg.com]
Linear Processes vs. Continuous Learning
The legacy workflow in most financial institutions is linear: application to maturity monitoring, with data collected once and frozen, plus manual handoffs and re-entry that limit scale.
The better frame is a cycle: origination, monitoring, policy refinement, and improved decisioning connected through data. Each interaction should make the next one faster and smarter. When institutions fail to retain and reuse verified data, they lose compounding learning, and they reset the clock on every deal.
Aggregation Is Not a Data Strategy
Aggregation centralizes views, but it rarely changes operational workflows. A true data foundation reduces redundant entry, enforces consistency, and injects trusted data into the workflow where decisions happen.
The Next Frontier: Persistent Borrower Intelligence
The persistent borrower profile
The future is persistent, cumulative borrower knowledge. Verified data should carry forward and improve over time, rather than being treated as disposable evidence for a single loan event. Persistence supports faster renewals, better risk assessment, and stronger relationships because the institution stops relearning the same borrower.
The new signals lenders are pursuing
Cash flow and transaction data: Datos Insights and Plaid report that 88% of U.S. consumer lenders have increased confidence using alternative credit data compared to a year ago, and 74% of consumers say they are comfortable sharing cash flow data with lenders. [4]
Alternative data beyond the bureau: LexisNexis reports 67% of lenders want more alternative data to approve more worthy borrowers, yet only nearly half currently have plans to expand along traditional scoring. Some examples lenders want include bank transaction data, employment data, payroll data, and utility data.
AI as an Orchestrator — Not a Layer
Embedded decisioning with human control
The goal is better decisions with fewer unnecessary touches. Embedded decisioning combines policy-aligned rules, automated checks, and workflow automation with explicit points for human override and auditability. Deloitte emphasizes that agentic systems raise operational, cybersecurity, privacy, reputational, regulatory, and legal risks, which makes governance and oversight non-negotiable. [5] [deloitte.com]
Agentic AI orchestration
Agentic AI is where systems move from answering to executing: interpreting objectives, breaking them into tasks, interacting with tools and systems, and adapting within constraints. [3] The World Economic Forum describes agentic AI as going beyond prompted generation to systems that can perceive, reason, act, and learn with less constant human guidance. [6] [mckinsey.com] [weforum.org]
McKinsey quantifies the upside when banks do this seriously: 3% to 15% higher revenue per relationship manager and 20% to 40% lower cost to serve, achieved by rewiring a domain end to end rather than layering AI on top of broken workflows. [3] [mckinsey.com]
KYA and the importance of agency
KYC and KYB matter when it comes to knowing the risks of your borrowers. When agents start taking actions on behalf of the institution, Know Your Agent (KYA) matters too: know the risk of the agent, not just the borrower. Design orchestration like an orchestra: specialized roles, defined scope, and subject matter expert control over the “composition.” Deloitte reinforces the same direction: embed compliance and oversight into the agent’s operating logic and monitoring mechanisms, not as a bolt-on later. [5] [deloitte.com]
What Leading Institutions Do Differently
- They attack high-friction handoffs like CRM-to-credit and decision-to-documentation, where time dies and risk hides.
- They prove value quickly in a defined domain and expand from real wins, not broad promises.
- They use data to reduce rework and risk, not just to build better dashboards.
- They scale accountable AI through segmented scope, explicit permissions, continuous monitoring, and subject matter expert-designed processes.
From Insight to Execution: Where Lending Leaders Move Next
Your institution has already invested in collecting and verifying borrower data.
You have already generated insights from portfolio performance.
But if your process still requires teams to:
- Re-collect data
- Re-validate information
- Reconcile inconsistencies
Then the challenge is not access to data.
It is the inability to operationalize it consistently.
This is where leading institutions are shifting focus — from visibility to execution.
They are building environments where:
- Borrower intelligence persists across the lifecycle
- Data flows directly into decisioning workflows
- Automation reduces friction without sacrificing control
- Consistency becomes a competitive advantage, not a compliance exercise
This is also where the right partner matters.
At Baker Hill, we help financial institutions move beyond fragmented workflows by embedding trusted, reusable data directly into the lending process — so teams can make faster, more consistent decisions with confidence.
Because the future of lending will not be defined by how much data you have.
It will be defined by how effectively you use it —
in every decision, across every relationship, at scale.
Sources
- KPMG, “Data modernization: Banks ready for the next stage” (2025 Banking Survey: Data Modernization, PDF). [kpmg.com]
- LexisNexis, “Global Lenders Turn to Alternative Credit Data to Detect Risk Earlier, Improve Portfolio Performance” (Mar 6, 2026). [lexisnexis.com]
- McKinsey, “Agentic AI is here. Is your bank’s frontline team ready?” (Dec 2, 2025). [mckinsey.com]
- Datos Insights and Plaid, “Cash Flow Underwriting: Reshaping the Lending Landscape” (Nov 2024 white paper PDF). [assets.ctfassets.net]
- Deloitte Center for Financial Services, “Agentic AI in banking” (Aug 14, 2025). [deloitte.com]
- World Economic Forum, “How Agentic AI will transform financial services” (Dec 2, 2024). [weforum.org]
