This white paper distils the webinar Single View of What: The Value of Data Matching in Entity Resolution into a practical reference for business and technical readers. It explains what entity resolution is, why organisations struggle to achieve a single view of customer, citizen, supplier, or counterparty data, and how data quality and data matching work together to solve the problem. It also summarises where machine learning and AI can add value, where human expertise remains essential, and how organisations should think about architecture, governance, and implementation.
Executive Summary
Many organisations talk about achieving a “single view” of an entity, but the practical reality is much harder than the phrase suggests. Customer, citizen, legal entity, and supplier records are often spread across multiple systems, created for different operational purposes, and maintained with varying levels of quality. This leads to duplicates, incomplete records, inconsistent formats, missing identifiers, and conflicting versions of the truth.
Entity resolution is the discipline of determining which records refer to the same real-world entity and then linking, clustering, or consolidating those records into a trusted view. The webinar emphasised that this is not just a technical matching exercise. Effective entity resolution depends on strong data quality foundations, careful standardisation, configurable match strategies, and input from subject matter experts who understand the data and its business context.
The value of a successful single view is substantial. Organisations can make better decisions, reduce operational inefficiency, improve service delivery, support compliance obligations such as KYC and AML, and identify risks including fraud and duplicate activity. In both financial services and public sector settings, the ability to connect fragmented data into a trustworthy entity view has direct operational and societal value.
The discussion also highlighted the role of AI and machine learning. These technologies can improve speed, recall, and classification of ambiguous matches, especially when deterministic rules alone are not enough. However, they should augment rather than replace governance, explainability, and expert review. Entity resolution remains a human-guided discipline supported by automation, not an unsupervised black box.
1. Introduction: Why “Single View” Matters
In data management and governance, the term single view is widely used: single customer view, single citizen view, single supplier view, and more. The webinar opened by challenging the simplicity of the phrase. A single view sounds straightforward, but in practice it means reconciling data that has been created in different places, for different reasons, at different times, and often with inconsistent standards.
Datactics described this challenge as common across industries. In financial services, firms need to align customer and counterparty data across systems to support compliance, onboarding, and risk management. In government and public services, organisations face the equivalent challenge with citizen data, vulnerable persons, case records, and service interactions. In each case, the goal is not simply to store data in one place, but to create a trustworthy, usable representation of the real-world entity behind the records.
The webinar positioned entity resolution as a core capability within broader data management: profiling data, detecting outliers, cleansing and standardising records, then applying matching logic to resolve duplicates and improve trust.
2. The Core Problem: Why Entity Resolution Is Difficult
Achieving a single view is difficult because enterprise data is rarely born clean. Records may be duplicated across systems, entered manually, updated at different times, or captured without a common identifier. Even when identifiers exist, they may be missing, outdated, or inconsistent across sources.
Common sources of complexity include:
- Variation in names, addresses, and descriptive fields
- Misspellings, abbreviations, and transposed characters
- Incomplete or missing attributes
- Different formatting standards between systems
- Conflicting versions of the same entity record
- Absence of universal identifiers across datasets
The transcript and related materials reinforce that matching is especially challenging for person and organisation data, where ambiguity is natural. For example, people matching must contend with naming conventions, incomplete personal details, and data quality gaps. Legal entity matching often depends on names, addresses, country information, and identifiers such as LEIs, which still require cleansing and standardisation before reliable matching is possible.
This is why the webinar repeatedly linked entity resolution to data quality. Matching poor-quality data directly tends to create false positives, false negatives, and high manual review effort. Better input data produces better resolution outcomes.
3. Entity Resolution Explained
Entity resolution is the process of identifying records that refer to the same real-world entity and deciding how they should be linked or merged. Depending on the use case, the outcome may be:
- Match: determine whether two or more records refer to the same entity
- Link: connect related records across systems without physically merging them
- Cluster: group multiple records believed to represent the same entity
- Golden record: create a trusted, canonical representation from multiple inputs
The supporting documentation describes a layered approach to this process. First, data is ingested and improved through cleansing and normalisation. Next, configurable match strategies are applied using combinations of exact, fuzzy, phonetic, numeric, regex-based, and semantic approaches. Finally, results are reviewed, scored, and either accepted automatically or escalated for expert validation.
This approach matters because no single match technique is sufficient for every domain. Exact matching may work well for strong identifiers; fuzzy or phonetic matching can help with spelling variation; NLP or semantic similarity can add value when unstructured or messy text is involved. Effective entity resolution combines these techniques rather than relying on only one.
Key principle: Entity resolution is best treated as a configurable, data-aware process, not a one-size-fits-all algorithm.
4. The Business Value of a Single View
The webinar made clear that entity resolution matters because it improves decisions and outcomes. When organisations can trust that records are correctly linked, they gain a clearer understanding of customers, citizens, suppliers, or counterparties and can act with greater confidence.
Key benefits include:
- Better decision-making: users can work from a more complete and accurate entity view
- Operational efficiency: less time spent reconciling duplicates or validating conflicting data
- Improved compliance: supports AML, KYC, customer due diligence, and regulatory reporting processes
- Fraud and risk detection: better visibility of suspicious duplication, hidden relationships, or fragmented records
- Service improvement: public sector organisations can reduce duplication and improve case visibility for citizen services
Internal reference material extends these examples. In financial services, matching supports single client view, anti-fraud measures, and cross-vendor mapping of instruments and identifiers. In the public sector, similar techniques are used for citizen data, case management, and vulnerable individual records. The same underlying challenge appears in different forms across domains.
5. A Practical Approach to Solving the Problem
The webinar presented entity resolution as a practical discipline rather than an abstract aspiration. Organisations should start by acknowledging that this is both a data and a business problem.
A sensible approach includes:
- Prepare the data: profile, cleanse, standardise, and normalise source records before matching.
- Design a match strategy: choose appropriate techniques for the entity type and attributes involved.
- Apply scoring and thresholds: use confidence scores across relevant attributes such as name, address, country, or identifier.
- Review ambiguous cases: route lower-confidence outcomes to trained analysts or domain experts.
- Refine over time: use expert feedback to improve rules, thresholds, and model behaviour.
This practical model is consistent with existing Datactics documentation, which describes flexible match strategies, scalable in-memory processing, and workflows that combine deterministic techniques with machine learning support.
6. Why Human Expertise Still Matters
One of the clearest messages from the webinar was that entity resolution is not a fully automated exercise. Sean McConnell described it as an art that requires input from multiple directions. Domain specialists such as crime registrars, analysts, or line-of-business experts understand the subtleties of their data and the consequences of getting a match wrong.
Technology and domain expertise must work together. Tools can accelerate matching, expose patterns, and reduce manual burden, but subject matter experts provide the context needed to judge edge cases, validate outcomes, and ensure the process reflects real business meaning.
Entity resolution requires the expertise of both subject matter specialists and the analysts and engineers configuring the tools. The balance between these perspectives is essential to achieving a trusted outcome.
This is especially important in high-stakes domains such as financial crime, citizen services, and vulnerable-person records, where false matches or missed links can have material consequences.
7. The Role of AI and Machine Learning
The webinar addressed AI directly and took a balanced position. AI introduces opportunities for greater speed, intelligence, and potentially improved accuracy, but also introduces risks, including overconfidence and poor explainability if applied carelessly.
Supporting material shows several areas where AI and ML can strengthen entity resolution:
- Classifying low-confidence matches when deterministic rules are inconclusive
- Learning from expert-reviewed match decisions to reduce future manual effort
- Improving semantic matching using NLP or embedding-based approaches
- Supporting data preparation tasks such as labelling, feature engineering, and standardisation
Datactics internal materials note that machine learning can reduce manual intervention significantly in some entity resolution workflows, and that low-confidence classifiers can learn from thousands of expert-reviewed examples. At the same time, strategy documents point toward newer embedding-based methods that can augment or replace older NLP approaches for better semantic matching.
Important: AI should augment governed matching processes, not replace them. Human review, auditability, and clear thresholds remain critical.
8. Architecture and the “Golden Record” Question
In the closing discussion, the webinar considered whether organisations should aim for a centralised architecture with one harmonised system of record. The answer was pragmatic: in an ideal world, deduplicated and controlled data would flow into a central platform, but real-world estates are more complex.
Entity resolution and architecture are not opposing choices. Matching can serve several architectural patterns:
- Creating a new golden record from multiple upstream systems
- Linking records across systems while leaving source systems intact
- Supporting a hub-and-spoke model where canonical identifiers are shared back to the estate
- Enriching governance platforms with trusted entity outcomes and trust scores
This is echoed in integration material involving Microsoft Purview, where Datactics performs advanced matching, clustering, and golden-record creation, then pushes canonical IDs and trust scores back into the governance layer. The implication is clear: a single view is as much an operational capability as it is a data architecture decision.
9. Use Cases Across Industries
| Industry or Domain | Example Entity Resolution Need | Value Delivered |
|---|---|---|
| Financial services | Single client view, KYC, AML, counterparty matching, instrument mapping | Compliance support, lower fraud risk, improved operational trust |
| Public sector | Citizen matching, service records, vulnerable individual visibility, case linkage | Better service delivery, clearer case context, reduced duplication |
| Enterprise data platforms | Cross-system golden record creation and canonical ID management | More reliable master data and downstream analytics |
These examples reinforce that entity resolution is not confined to a single vertical. The data patterns differ, but the core challenge remains the same: connect fragmented records accurately enough to support real decisions.
10. Conclusion
The webinar’s central argument was simple: a single view is valuable, but it is not achieved by wishful thinking or architecture alone. It requires disciplined data quality, flexible match strategies, domain expertise, and controlled use of AI where it adds measurable value.
Entity resolution is foundational for organisations that need to trust their data across systems. Whether the goal is a single customer view, citizen view, supplier view, or legal entity view, the path forward is practical: improve the data, match it intelligently, involve experts, and operationalise the result.
For organisations dealing with fragmented and duplicate entity data, the opportunity is not just cleaner records. It is better decisions, stronger compliance, improved service, and a more trustworthy data foundation overall.
Next Steps
You can watch a full recording of the webinar below. Need more information? Simply hit ‘Get in touch’ at the top of the page.