The Path to Modern MDM: Client Domain Data Quality

Our team at Alphyn.ai has spent many years implementing and adapting various IT solutions, including MDM systems. Combining that accumulated knowledge, we released our own high-performance multi-domain product, Alphyn Governance MDM.

We strove to make Alphyn Governance MDM a flexible, comfortable, and performant solution, and carefully studied the market and its needs. Today we want to share the analysis results and reflect on why MDM solutions are needed by modern business, what role they play, and what tasks they address -- using the example of the client data domain.

The Significance of Client Data Quality for Business

Companies beginning to implement a "Know Your Client" (KYC) program often underestimate the importance of client data quality. When the number of sources in an organization is still not large, a decision is typically made to entrust client data management to one of the Systems of Record (SOR) -- for example, an oCRM.

A single client data entry system can satisfy the organization's needs for quite some time. But this approach loses relevance as new client data sources and enrichment sources appear.

As new data sources, client information replicas, and systems with enriched data appear, maintaining the quality of client data becomes more difficult. Business processes around source systems are typically isolated, so quality problems may go unnoticed for a long time and accumulate. At this stage, an attempt may be initiated to bring order and consolidate all available client data in a single data mart by developing deduplication scripts.

Such solutions are a half-measure and typically stagnate, while the systemic solution to the problem is correctly addressed by implementing an MDM-class product. The stimulus for consolidation often becomes processes that are heavily dependent on client data quality. These may be CVM initiatives, AML development, or Data Science projects.

If the data strategy does not account for the necessary organizational changes -- personnel and process transformations -- while focusing exclusively on transformation through IT solution implementation, and if MDM implementation is given only secondary importance, then these projects risk encountering serious problems or even failure.

Custom Development vs. Ready-Made Solution: Requirements and Challenges

The model described above can be realized in various ways:

• Custom development from scratch
• Development based on one of the existing solutions in the organization
• Implementation of a ready-made MDM solution

The first two approaches are associated with certain risks and the inherent inflexibility of technologies. When creating a custom analytical MDM, the client will inevitably encounter the most common problems:

• A UI will most likely be absent. The mechanism's operation will have to be analyzed at the database level.
• The data model will be administered at the database level.
• Change history storage and logging functionality will probably be excluded for cost savings.

Costs for such a solution may greatly exceed the initially allocated budget.

Strategically it will be more correct to focus on implementing a professional MDM solution, which must possess the following characteristics:

• Automated golden record formation process -- The user must be able to participate. But the less frequently employees intervene in the system's operation, the greater effect it can deliver. This is especially important to consider when source systems contain tens of millions of client records.

• Business logic for working with the required domain -- It is important that the client can make changes without vendor participation. Business continuously evolves, and while the rules for correcting typos or parsing addresses according to a standardized address register change rarely, the rules for classification or name parsing according to dictionaries need to be detailed several times a year. Many solutions on the market offer business logic customization, but this makes no sense if it has to be developed from scratch over months.

• Ability to fine-tune and expand the data model -- Continuous work on data quality and connecting new sources to the system require that the data model can change. The system must be able to track which version different records are stored under and possess tools for upgrading versions. This is especially important when connecting large arrays of external data and adding fields for storing derived values.

• Horizontal scalability and the ability to operate in high-performance operational mode -- Installing MDM is only the first step toward understanding the organization's data. And according to the data strategy activity plan, preparation for implementing real-time client data consumers is best begun several months before the start of the corresponding projects. At this stage there is usually no need to deploy the MDM solution in a resource-intensive operational mode, but after performing several quality improvement cycles and as consumer implementation projects are completed, such a necessity will arise.

Alphyn Governance MDM as a Ready-Made Business Solution

Our team at Alphyn.ai develops IT products for business, including in the field of master data management. The product line includes a high-performance solution for cleaning, enriching, standardizing, and deduplicating client data -- Alphyn Governance MDM -- which is useful for medium and large businesses developing the "Know Your Client" principle. The tool meets all the principles listed in the previous section.

Work on data quality cannot be limited to the immediate implementation of a system. This is the continuous execution of research and corrective activities meeting the organization's needs. The system must be sufficiently flexible and adaptive to support their execution.

In Alphyn Governance MDM there are several levels for focusing team attention on the data quality improvement process:

1. At the first level are quality codes -- the most numerous entity. They accompany every piece of information passing through the solution. Business logic rules triggered over a record mark a field, a group of fields, or the entire record with the result of their work.

   Some codes mark fields where the system made changes independently; others mark cases where automatic changes are impossible and human attention is required. If the problem corresponding to a code is corrected and the code is no longer valid, it will automatically be removed -- but the information that the code was once set will remain in the history permanently for retrospective analysis and statistics review.

2. When it becomes clear that a quality code or group of codes requires special attention, a rule for creating incidents is configured. In Alphyn Governance MDM, incidents are not created by default for every event found by the system, so as not to overload user attention. Incidents are created and automatically assigned to responsible parties according to predefined logic. If the cause of an incident's occurrence is eliminated, it will be automatically closed with the date and time recorded. Each user sees the incidents assigned to them and can focus on resolving them.

3. For proactive work on data quality, there are visualization, search, and filtering mechanisms that allow selecting particularly important data quality violations, forming planned activities from them, and distributing them among responsible employees.

Final Thoughts

The rules for forming master records are individual for each organization. Data quality in sources can vary. Some systems may be more trusted than others, and in some cases the decision logic for which records should provide values for master fields can be even more complex. Incident review and proactive data investigation often lead to the need to adjust master formation logic: adding handling for specific cases.

At this point it is important that the system has tools for automatic re-formation of master records accounting for possible changes in merges and splits, and handles unique client identifiers with maximum care. This determines how organically the changes will be accepted by other systems in the organization's IT landscape subscribed to the reverse data flow from MDM. Alphyn Governance MDM was originally designed to take maximum account of all possible aspects of integration.

Bringing order to data is no easy task, and it is important to build this process with a modern, flexible, and high-performance tool like Alphyn Governance MDM.