What is master data management?

Master data management is the set of rules and connections by which a company keeps one record per customer, supplier, product and site across its systems. It decides what counts as one of them and which system holds the leading version of each field. The short form is MDM.

What master data are

Master data are the records that other records point to: customers, suppliers, products and sites. An order points to a customer and to a product. The order itself is transactional data, because it records one event. The customer record stays in use for years. It changes when the customer moves or takes a new name.

Why a company needs master data management

In a growing company these records tend to appear system by system. The CRM knows a customer under one number and the ERP under another. The shop knows the same customer under an email address, and accounting under a debtor number. Each system keeps its own copy of the address, and each copy changes on its own. A simple question then has several answers. One example is the number of active customers.

What master data management defines

The first result is a definition per type. It says what counts as one customer, one supplier or one product in this company.

The second result is a field table. It names the leading system for each field of a record. The ERP can lead for the legal name and the billing address, and the CRM for the contact persons. When two systems disagree, the value of the leading system wins. The other value stays in the history with its source. The record that the other systems follow is often called the golden record.

The third result is the matching of records, also called entity resolution. It works with keys that several systems share. A VAT ID, a commercial register number and an article number exist outside the company’s systems. Registers can reuse a code. So a shared key proposes a match, and a check decides it. A contradiction blocks the match.

When two records describe one company, a link says so. Both records keep their numbers and their history, and a wrong link comes off in one change. The strict form of such a statement is owl:sameAs in the OWL 2 specification of the W3C. The number a system gives a record is a handle. Object identity is the question of which thing in the world the record describes.

Does a company need MDM software?

A company can start without it. The definitions and the field table come first. They show whether a dedicated tool pays off. One of the existing systems can take the lead, for example the ERP for customer numbers.

Master data management, data quality and data migration

Master data management settles which record leads and how the systems share it. Data quality work cleans the records inside one system. A data migration moves records from an old system into a new one. Duplicates that turn up in its field mapping are a question for master data management.

My work with master data

I am an IT expert for data platforms. I have designed data models for more than twenty years. For an insurance group I connected a customer portal for insurance policies to an existing CRM system. I have helped build a client’s information platform. On it, companies and people keep one identity across registers and sources.

In a master data project the other systems write through an API, with an access token per system. An import token may create records and may not change existing ones.

Today AI agents speed up how I build and transform data platforms. I build software with Claude and Codex. On the client’s platform, AI agents review the pairs of records that rules cannot settle. When an agent is unsure, both records stay.

In master data management consulting I write the definitions and the field table together with the people who know the records. I take on this work as a freelancer or as a permanent employee.