“Single Source of Truth” is one of the most widely used concepts in banking technology and data management. In theory, it promises consistency, transparency and better decision‑making.
In practice, however, many banks still struggle to achieve it, even after significant investment in data platforms, integration programmes and system modernisation.
So why does it keep failing?
The Promise of a Single Source of Truth
The idea is simple: one authoritative data foundation where all balance sheet, risk, liquidity and customer data is consistent and trusted across the organisation.
In banking, this would ideally mean:
- One view of the balance sheet
- One set of risk metrics
- One consistent basis for ALM, treasury and regulatory reporting
This promise is attractive because fragmented data environments clearly cause inefficiencies and inconsistencies. Siloed systems often lead to duplicated effort, mismatched data and delayed decision‑making.
But the gap between theory and reality remains significant.
Why It Fails in Practice
1. Data Is Inherently Distributed in Banks
Banks do not operate on a single system.
Core banking, loan origination, risk engines, treasury platforms, compliance systems and regional systems all generate and own critical data. These systems evolved for specific purposes and were never designed to operate as one unified model.
Attempting to force everything into a single physical source often introduces complexity, latency and operational fragility rather than clarity.
The result is not one “truth”, but multiple versions of reality that need constant reconciliation.
2. Data Silos Persist Despite Integration Efforts
Even after modernization programmes, data fragmentation remains a core challenge.
In many organisations:
- New systems are added alongside legacy platforms
- Departments optimise locally rather than globally
- Integration happens incrementally rather than structurally
Over time, this creates layered architectures where modern and legacy systems coexist without consistent data pathways.
3. Definitions Are Not Aligned Across Functions
In banking, the issue is not only technical, it is semantic.
Key concepts such as:
- Customer
- Exposure
- Liquidity
- Net interest income
can have different definitions across risk, finance and treasury functions.
Even with integration in place, inconsistent definitions lead to conflicting outputs and reduce trust in reported figures.
From a decision‑making perspective, this is critical: you cannot have a “single truth” if the underlying concepts are not shared.
4. Centralisation Creates Bottlenecks
A fully centralised data model often introduces operational friction.
When all data changes, transformations and reporting requests must go through a central team:
- Delivery cycles slow down
- Business teams lose flexibility
- Workarounds emerge
In reality, teams frequently revert to spreadsheets and local datasets when the central system cannot meet their needs quickly enough.
Ironically, the attempt to enforce a single truth often creates multiple unofficial ones.
5. Manual Reconciliation Never Disappears
Many banks still rely on manual reconciliation between systems before producing management or regulatory reports.
Different systems generate different outputs for the same business reality, requiring:
- Parallel data pipelines
- Adjustments outside core systems
- Hidden reconciliation processes
This creates a situation where leadership sees the final numbers, but not the effort required to produce them, or the underlying inconsistencies.
This is one of the clearest signs that a true single source of truth has not been achieved.
6. Governance Is Often Underestimated
A single source of truth is not just a technology problem, it is a governance problem.
It requires:
- Clear ownership of data
- Consistent definitions
- Strong controls over data quality and usage
Without a robust governance framework, data may be centralised but still inconsistent, outdated or unreliable for decision‑making and regulatory reporting.
The Deeper Issue: “Truth” Is Contextual
One of the fundamental challenges is that “truth” in banking is rarely universal.
Different functions require different perspectives:
- Treasury focuses on liquidity and funding
- ALM focuses on structural risk and long‑term projections
- Risk focuses on exposure and capital
Each view is valid within its context, but not identical.
Trying to force all perspectives into a single rigid model can reduce usability and lead to loss of business context, ultimately undermining trust.
What Works Better in Practice
Rather than pursuing a strict single source of truth, many banks are moving toward more pragmatic approaches.
A Governed Single Version of Truth
Data may remain distributed, but:
- Definitions are aligned
- Calculations are consistent
- Outputs are reconciled and traceable
Integrated but Not Over Centralised Architectures
Systems remain specialised, but:
- Data flows are controlled and transparent
- Key metrics are derived consistently
- ALM and treasury views are aligned on the same balance sheet logic
Focus on Decision Support
The goal shifts from asking where the data is stored to asking whether decisions are consistent, explainable and aligned across functions.
What This Means for ALM and Treasury
For ALM and treasury specifically, this has direct implications.
A technically centralised system does not guarantee consistent balance sheet views. Integration must include assumptions, models and definitions, not just data feeds. Decision support requires aligned analytics, not just shared data.
This is why combined ALM and treasury platforms focus on:
- A shared balance sheet representation
- Consistent scenarios and assumptions
- Direct linkage between treasury actions and ALM outcomes
Final Thoughts
The “Single Source of Truth” remains a useful concept, but in many banks it is still more aspiration than reality.
The reasons are not purely technical:
- Data is naturally distributed
- Definitions differ across functions
- Systems evolve independently
- Governance is complex
The result is that many banks operate with multiple versions of reality, reconciled through process rather than design.
For institutions focused on better decision‑making, the more practical goal is not a single system, but consistent, transparent and governed data across ALM, treasury and risk.