What is MDM?

Master data management explained around the problem, not the product category.

The problem

Your company has the same customer in eight different systems. The CRM has them as "Acme Corp". Stripe has them as "Acme, Inc.". The support tool has them as "Acme Corporation, LLC". Marketing automation has two records — one with a typo. The data warehouse has all four. Nobody actually agrees on what the customer's annual revenue is, which contact is the decision-maker, or how many open support tickets they have.

This is normal. It happens because every system was bought, built, or evolved independently. Nobody designed it to be inconsistent — it just is.

Master data management (MDM) is the discipline of producing one trusted version of each customer (or product, or vendor, or any business entity) from those scattered sources.

What "one trusted version" actually means

It doesn't mean "pick a system and trust it". That only works if your business has exactly one system, which it doesn't.

It means producing a golden record for each real-world entity — a synthesized row whose field values come from rules you can inspect and change. The golden record links back to every contributing source row, so you can answer:

• Which sources said this customer exists?
• Why was the address from Stripe chosen over the one in the CRM?
• When was each field last updated, and by which source?
• Are there other potential duplicates we haven't merged yet?

Without lineage like this, "trusted" is just a vibe. With it, "trusted" means "auditable".

Why it's hard

Three things make MDM more work than it looks:

1. Matching is fuzzy. "Acme Corp" and "Acme, Inc." are probably the same company, but they're not the same string. The algorithm that decides has to deal with typos, formatting differences, missing fields, and the occasional case where two different companies share a name. See Entity resolution.

2. Field values disagree. Once you've decided five rows are the same customer, you still have to pick one phone number, one address, one annual-revenue value. Different fields need different rules — "most recent" for contact info, "highest confidence source" for legal name. See Survivorship.

3. The data keeps changing. New rows arrive every day. Sometimes they're brand-new customers. Sometimes they should merge into an existing golden record. Sometimes a previous merge turns out to have been wrong. MDM is a continuous process, not a one-time export.

Where Golden Suite fits

Golden Suite is the engine for steps 1 and 2 plus the surface for step 3. You bring sources; Golden Suite produces and maintains the golden records, with the lineage that lets you defend every decision.

The rest of these concept pages drill into the specific moving parts.