This guide is for CTOs, heads of data and operations leads at mid-market companies who are being told they need master data management (MDM) and want to know whether that is true yet. It sits under Netbase's data engineering and platforms service and pairs with the guides on data catalog and lineage for a 20-person data team, on change data capture for operational databases and on ERP, CRM and e-commerce integration architecture, which decides which system owns each record.
In this guide
- What problem does master data management actually solve?
- Which triggers justify master data management?
- What are the cheaper steps before MDM?
- Registry, consolidation, coexistence or central hub?
- When should a company not buy an MDM tool?
- How to move from spreadsheets to managed master data
- Where AI fits
- What delivery record exists, and what does not
- Alternatives: other ways to get one version of a record
- Limits of this guide
- Frequently asked questions
- Next step
What problem does master data management actually solve?
Master data management solves one problem: the same real-world customer, product, supplier or location is described differently in several systems, and the business cannot say which version is right. An MDM tool matches those records, decides which value wins for each field, keeps the result as a golden record and gives people a workflow to review doubtful matches.
It does not fix a missing process. If nobody is allowed to say which system owns a customer's billing address, an MDM tool will match the records and then wait for a decision that never comes. The tool automates a policy; the policy has to exist first.
Most mid-market companies feel the problem in three places: duplicate customers in the CRM and the ERP, product data that differs between the ERP and the online store, and supplier records copied by hand between purchasing and finance. Each can often be solved without a dedicated MDM platform.
Which triggers justify master data management?
The triggers that justify MDM are about how many systems change a record and what a wrong record costs, not about row counts. The matrix below lists six common triggers, the cheaper response that often removes each one, and the point at which an MDM tool becomes the better answer.
-
Several systems edit the same record
- What you see
- Address or terms changed in the CRM but not in the ERP
- Cheaper response first
- Name one owner system per field and sync one way
- MDM becomes worth it when
- Two or more systems must keep editing the same fields
-
Duplicates reach money or customers
- What you see
- Two accounts, double invoices, split order history
- Cheaper response first
- Unique keys, a duplicate check at entry, a weekly duplicate report
- MDM becomes worth it when
- Duplicates keep returning from imports, partners or acquisitions
-
Reports disagree on counts
- What you see
- Sales and finance quote different customer or product totals
- Cheaper response first
- One shared customer and product table in the warehouse
- MDM becomes worth it when
- The fix must also flow back into operational systems
-
Hand-written matching rules keep growing
- What you see
- Dozens of exceptions in spreadsheet formulas or SQL
- Cheaper response first
- Probabilistic matching run in the pipeline, with a review list
- MDM becomes worth it when
- Matching needs stewards, an audit trail and a review workflow
-
Hierarchies matter
- What you see
- Parent companies, product families, ship-to and bill-to accounts
- Cheaper response first
- Hierarchy tables owned by one team
- MDM becomes worth it when
- Hierarchies change often and drive pricing, credit or reporting
-
An audit or regulator asks who changed what
- What you see
- No history of record changes
- Cheaper response first
- Change history and access control in the owner system
- MDM becomes worth it when
- History must span several systems for the same record
One trigger alone is usually handled by the cheaper response. When three or more apply to the same domain, such as customers, an MDM tool deserves a proper evaluation.
What are the cheaper steps before MDM?
The cheaper steps are ownership, identifiers and checks. Together they often remove most of the pain people buy MDM to remove, and none of them is wasted if MDM comes later, because every MDM rollout needs the same decisions.
- Name an owner system per field. Write down, for each customer and product field, which system may change it and which systems only receive it. This decision is the core of the integration design, not a separate project.
- Keep a cross-reference of identifiers. Store the ERP, CRM and store identifiers of each record side by side, so a record can be matched by key instead of by name.
- Check at entry. Search for a likely duplicate before a new customer or supplier is created, in the owner system.
- Check in the pipeline. Uniqueness and not-null tests on key columns, plus a duplicate report that a named person works through each week, catch what slips past entry checks.
- Build one shared table for analytics. A deduplicated customer and product table in the warehouse gives reports one count, even while the operational systems still hold duplicates.
- Use probabilistic matching where rules stop scaling. Open-source libraries such as Splink compare several fields and score how likely two records describe the same entity, without unique identifiers or labelled training data.
Registry, consolidation, coexistence or central hub?
| Option | Strength | Risk | Choose it when |
|---|---|---|---|
| Registry or consolidation (an index or a cleaned copy) | Little change to source systems; fast to start | Source systems keep their duplicates; fixes do not flow back | Reporting and analytics need one view; operations can live with the duplicates |
| Coexistence (golden record synced back to sources) | Clean records reach operations without replacing systems | Two-way sync and conflict rules are hard to get right | Several systems must keep editing records and duplicates cost money |
| Central hub (records created and edited only in the hub) | Strongest control, audit trail and consistency | Every application must change; longest rollout | Regulation or audit demands it, or record creation is already centralised |
Public guidance from MDM vendors such as Informatica describes the same progression: many organisations start with a registry or consolidation style for one or two domains, then move to coexistence or a central hub as their governance matures. Start with the domain where a wrong record costs the most.
When should a company not buy an MDM tool?
Do not buy an MDM tool when one system already owns each record and the others only read it, when the problem lives inside one department or application, or when nobody will act as a steward for doubtful matches. A tool without stewards produces a growing queue of unreviewed matches, and the old spreadsheet comes back beside it.
Also wait when the record set rarely changes: a short list of warehouses or legal entities is better kept in one table with an owner than in a platform. Re-check the trigger matrix when a new system, an acquisition or a new sales channel arrives, because each adds a system that edits the same records.
How to move from spreadsheets to managed master data
-
Pick one domain
Choose customers, products or suppliers, whichever causes the most rework or the costliest errors today; do not start with all three.
-
Map who edits what
List every system and team that creates or changes a record in that domain, field by field.
-
Decide the owner per field
Agree which system owns each field, and change the integrations so the other systems receive the value instead of editing it.
-
Add identifiers and entry checks
Store each system's identifier on the record and add a duplicate search before a new record is created.
-
Measure duplicates in the pipeline
Add key tests and a weekly duplicate report with a named owner, and track whether the report shrinks.
-
Run matching where rules fail
Apply probabilistic matching to the records the rules miss, and send doubtful pairs to a person for review.
-
Evaluate MDM against the triggers
If three or more triggers still apply after these steps, evaluate an MDM tool with the owner decisions, identifiers and match rules you already have as its starting configuration.
Where AI fits
AI helps most with the matching and review work around master data. A model can suggest likely duplicate pairs that rules miss, standardise free-text fields such as company names or product titles, and draft a short reason for each suggested match so a reviewer decides faster. Netbase works with the major commercial and open-source AI tools and models, chosen per project.
People still decide. A steward approves every merge that touches money, credit or contracts, engineers set the match thresholds and test them on known pairs, and no model deletes or merges a record on its own.
- In delivered work
MLOps pipeline
An MLOps pipeline among the AI projects Netbase has delivered for clients that are not named; an AI delivery record, not an MDM rollout.
- Available capability
AI and data services
AI-suggested matches with human review are a way Netbase approaches the work, not yet tied to a published case.
What delivery record exists, and what does not
Netbase's ERP work covers Odoo consultancy and customization, cloud ERP platform development and e-commerce integration with ERP through API sync of product catalogues, inventory, orders and customer data. In that work, deciding which system owns a customer or product record is part of the integration design.
Netbase has delivered AI projects for clients that are not named, including MLOps pipelines, and its named AI delivery is the 4over4 recommendation engine. Delivery runs in a six-step lifecycle from discovery to ongoing support, and most projects are delivered on fixed-price contracts. No published Netbase record describes an MDM platform rollout, and no duplicate rate, cost or time figure is claimed here.
Alternatives: other ways to get one version of a record
- The ERP as the owner. When the ERP already holds customers, products and suppliers, make it the only system that creates them and sync outwards; choose it when other systems can accept read-only copies.
- A product information management (PIM) system. For rich product content across many channels, a PIM owns product descriptions and media while the ERP keeps price and stock; choose it when the product domain is the pain.
- A customer data platform. A customer data platform unifies customer profiles for marketing; choose it when the need is segmentation and campaigns, not billing or credit records.
- A warehouse-only golden table. Deduplicate in the warehouse for reporting and leave operations as they are; choose it when the business problem is disagreeing reports.
Limits of this guide
- The trigger matrix is a planning aid, not a benchmark; regulated industries may need MDM features earlier.
- No MDM product is recommended here; the right tool depends on your domains, your systems and who will steward.
- Matching quality depends on your data; test any match rule or model on pairs whose answer you already know.
Plan the next step with a Netbase consultant
Frequently asked questions
You need it when several systems keep editing the same customer, product or supplier records, duplicates keep returning after clean-ups, and the fix must reach operational systems, not only reports. One of these alone is usually solved by ownership rules and pipeline checks.
For one domain and a few sources they often can, especially with probabilistic matching and a weekly review list. They stop being enough when matches need stewards, an audit trail and changes written back to several systems.
A golden record is the single trusted version of a customer, product or supplier, built by matching duplicates and choosing the winning value for each field. It is only as good as the ownership rules behind it.
Start with the domain where a wrong record costs the most, usually customers when invoices and credit are at stake, or products when several sales channels show different data. Prove the approach there before adding a second domain.
Next step
If duplicates and disagreeing records are costing your team time, Netbase can map who edits which fields, propose the owner per field and the cheaper steps, and say whether an MDM tool is justified yet: book a solution review.
Related services and solutions
Data engineering platforms for reporting and AI Netbase builds data pipelines and analytics foundations for companies whose data is spread across a store, an ERP, a CRM and spreadsheets. We collect it, clean it, model it and check its quality, so reports agree with each other and AI projects start from data people trust. Data engineering is a capability we are growing alongside the platforms we deliver.
Learn More
API integration services that keep your systems in agreement and ready for AI Netbase provides API integration services for commerce and operations teams that need e-commerce, ERP, CRM, shipping and payment systems to agree, so data moves once and AI services can act on it, with retries, monitoring and reconciliation on every API and webhook. According to its published client record, Netbase has worked with 500+ businesses across 80 countries since 2012.
Learn More
Enterprise integration hub ready for AI agents: one monitored layer for commerce, ERP and CRM data An enterprise integration hub is a central layer that moves orders, customers, stock and invoices between your commerce, ERP and CRM systems, with monitoring, retries and replay in one place, and scoped APIs that AI agents can call safely. Netbase builds it as a custom layer on the stack you already run, starting from the flows that break most often.
Learn More
Work with Netbase remotely
Every Netbase engagement is delivered remotely, so clients in any country and any industry can work with us; Netbase has worked with clients in 80 countries. Contracts and invoices are issued by Netbase JSC, and payment is made by bank transfer to Netbase JSC's company account or PayPal Business. Talk to the CEO, Huy Nguyen (David), on WhatsApp or by email.
Discuss a project
Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.project@netbasejsc.com
+84 937 869 689
91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam
Get in touch
Tell us what you want to build, modernize, or operate.