Customer Master Data

A distributor enters the same buyer twice under slightly different names, so one account receives invoices on thirty-day terms while the other has a separate credit limit. The duplicate weakens the total view of exposure and may confuse collections.

Master data describes who the customer is and how routine transactions should work, which differs from a single order or invoice. Use a unique identifier and verify the legal entity, because trading names, branches and group companies may need different treatment, and a mistaken merge can combine unrelated buyers while unnecessary duplicates fragment the relationship.

Set fields according to the process: billing addresses, delivery locations and tax details serve different purposes, so a warehouse needs the delivery point while finance may need the correct legal invoicing details. Do not assume every customer record should contain bank details; they may be relevant for refunds or direct-debit arrangements, but collecting them without a purpose increases risk, so separate sensitive payment information and restrict its use.

Define who may create a customer and what evidence is needed, such as a legal name, contact, tax information or an approved sales relationship, and search for duplicates before creating another account. Compare names, registration numbers, addresses and related entities rather than relying on an exact-text match, and investigate apparent matches because two entities can share a similar name.

SAP's master-data governance learning material describes customer models, workflows, duplicate checks, validation and change histories, which are software features, not proof that every business needs a particular platform, and IBM describes master data management as connecting trusted, consistent core data across applications, a principle that is useful even when a business uses a small number of systems. Assign data stewardship, since sales may own contact updates, finance may own terms and tax fields, and credit may approve limit changes, and a shared record should not mean every user can rewrite every field.

Require approval for consequential changes: raising a credit limit changes risk, altering payment terms changes cash timing and changing a bank destination may enable fraud, so log who requested, reviewed and applied the change. Check high-risk fields against an independent channel, since an email asking to change refund details should not automatically override a verified record.

Link records between systems carefully, as the CRM may store relationship notes while the ERP owns invoicing settings, and synchronisation rules should say which system wins when details differ, or errors will keep reappearing. Treat inactive customers clearly by retaining records as long as required for legitimate accounting or legal purposes while controlling access and deleting data under applicable policies.

Protect personal data inside the customer file, since contact names, phone numbers and transaction history may be regulated, limit access by role and document what is retained and why. Data quality has several parts: completeness, accuracy, uniqueness and timeliness, and a record can be complete but wrong, or correct last year but stale today, so review the fields that drive the most consequential transactions first.

An illustrative audit might sample 500 customer records and find 470 free of the defined errors, a sample pass rate of 94% that is not necessarily the accuracy rate of all customers, especially if the sample omitted old or high-risk accounts. Use exception reports such as returned invoices, bounced emails, duplicate tax IDs or conflicting credit terms as prompts to investigate, fix the source record as well as the immediate transaction, and establish one accountable route for corrections rather than arguing which spreadsheet is right; for owners, accurate master data reduces preventable friction but does not by itself eliminate bad debts or failed deliveries, so measure improvement in those outcomes without claiming the cleanup caused every change.