Master Data Management
Master Data Management is the discipline of defining, owning, and synchronizing the most important business entities — customers, products, suppliers, employees, locations, accounts — across all systems. The goal is a 'golden record' for each entity: one authoritative version of the truth that all systems trust. Without MDM, the same customer exists 7 times across CRM, ERP, billing, and support, with different names, addresses, and IDs. Marketing emails them twice; finance bills them at the wrong address; support can't find their history. MDM is unsexy plumbing that quietly determines whether your data, AI, and customer experience can ever work. Forrester estimates MDM problems cost large enterprises 15-25% of revenue in inefficiency and lost trust.
The Trap
The trap is treating MDM as an IT data-cleansing project. Buying an MDM tool (Informatica, Reltio, Stibo) and pointing it at messy data without resolving the political question 'who owns customer master?' produces a beautifully integrated golden record that nobody trusts because no business owner stands behind it. The other trap is scope creep: trying to master 40 entities at once. Successful MDM programs master 1-3 high-value entities (almost always customer first) and prove value before expanding. 70% of enterprise MDM programs are over budget or abandoned within 3 years — almost always due to ownership ambiguity, not technology failure.
What to Do
Pick ONE master entity (usually customer) with a clear executive sponsor (usually the CRO or COO). Establish a single business owner with authority to arbitrate disputes. Define the survivorship rules (when CRM and ERP disagree on the customer's address, who wins?). Build the golden record incrementally: start with 5-10 critical attributes, not 200. Measure success by downstream KPIs: duplicate rate, match rate, time-to-onboard a customer, marketing send waste. Only after this works for one entity, expand to product or supplier.
Formula
In Practice
Coca-Cola spent ~5 years (2014-2019) building a global Customer MDM after realizing they had ~70 different definitions of 'customer' across regions and bottling partners. Same retailer (e.g., Walmart) was represented as 8 different entities in different systems. Outcomes after the program: 30% reduction in trade-promotion waste (no longer paying twice for the same shelf space), 15% improvement in account-level forecast accuracy, ability to negotiate global contracts with global retailers because they finally knew their global revenue per customer. The program required dedicated business owners per region with veto power over local definitions — the political work was harder than the technology.
Pro Tips
- 01
The hardest MDM question is usually political: when CRM says the customer's billing address is X and ERP says Y, who wins? The answer must be a business rule, not 'most recent timestamp'. Without explicit survivorship rules signed by the business owner, your golden record is just an opinion.
- 02
Start with customer master; everything else (product, supplier, location) is secondary. Customer is where ROI is highest (revenue, marketing, CX) and where business sponsorship is easiest to get. Don't let IT push you toward product master first because it's 'technically simpler' — that's a graveyard.
- 03
Build a Data Stewardship layer: 5-15 trained humans who manually resolve the 5-10% of records the automated matching can't handle. MDM tool vendors will tell you matching is 99% automated; in practice, the messy 5% drives 80% of the value (your largest customers, your most complex accounts).
Myth vs Reality
Myth
“Modern AI/ML eliminates the need for MDM”
Reality
AI makes MDM more important, not less. ML models trained on duplicate, conflicting customer records produce conflicting predictions. The garbage-in-garbage-out problem amplifies at AI scale. Companies investing in AI without MDM build expensive engines that recommend wrong things confidently.
Myth
“A CDP (Customer Data Platform) is the same as Customer MDM”
Reality
CDPs are marketing-focused: they unify behavioral data for activation. MDM is enterprise-wide: it serves finance, ops, support, legal, and marketing. CDPs typically don't have the survivorship rules, stewardship workflows, or integration depth required for invoice accuracy or regulatory reporting. Many companies buy a CDP and discover they still need MDM for the rest of the business.
Try it
Run the numbers.
Pressure-test the concept against your own knowledge — answer the challenge or try the live scenario.
Knowledge Check
A bank has 12M customer records across 15 systems. Their MDM program produces a 'golden record' but the new Chief Marketing Officer says the data 'doesn't feel right' and refuses to use it. The technical team insists the matching algorithm is 96% accurate. What is the most likely root cause of the failure?
Industry benchmarks
Is your number good?
Calibrate against real-world tiers. Use these ranges as targets — not absolutes.
Customer Duplication Rate (Pre-MDM)
Mid-to-large enterprises across CRM, ERP, billing, support systemsExcellent (well-governed)
< 5%
Good
5-12%
Average Enterprise
12-22%
Severe
22-35%
Crisis
> 35%
Source: https://www.forrester.com/report/the-forrester-wave-master-data-management/
Real-world cases
Companies that lived this.
Verified narratives with the numbers that prove (or break) the concept.
Coca-Cola
2014-2019
Coca-Cola embarked on a multi-year global Customer MDM program after discovering the same retailer (e.g., Walmart) was represented as 8+ different entities across regions and bottling partners. They invested in survivorship rules, regional data stewards, and a global customer hierarchy linking parent companies to local stores. The program required executive sponsorship at the CCO level and named regional data owners with authority over local exceptions. Outcomes: 30% reduction in trade-promotion duplication, 15% better forecast accuracy, ability to negotiate global retailer contracts with confidence.
Customer Definitions Pre-MDM
~70 across regions
Trade Promotion Waste Reduction
~30%
Forecast Accuracy Improvement
+15%
Program Duration
~5 years
Global MDM is a 5-year, executive-sponsored program. The political work (regional data stewards, arbitration authority) is harder than the technology, and it is what unlocks the value.
Procter & Gamble
2010-present
P&G runs one of the most sophisticated enterprise MDM environments globally, with dedicated data governance councils for customer, product, and supplier domains. Their product master alone covers ~10,000 SKUs across categories with strict naming, hierarchy, and attribute standards. The MDM platform is the backbone for retail-link analytics with major retailers (Walmart, Target, Kroger) — without it, P&G could not negotiate with confidence on category-level performance.
Master Domains Managed
Customer, Product, Supplier, Location
Product SKUs in Master
~10,000 globally
Data Governance Councils
Multiple, by domain
Annual MDM Investment
Estimated $20M+
Sustained MDM is an operating capability, not a project. P&G's competitive advantage in retail analytics rests on master data quality that took 15+ years to build.
Hypothetical: $4B Healthcare System
2019-2022
A regional hospital system bought a top-tier MDM platform ($6M license + $4M services) to unify patient records across 12 facilities. After 30 months, the platform produced golden records but clinicians refused to trust them — there was no medical director who had publicly endorsed the matching rules, and several near-miss incidents involving merged records of different patients with the same name created legal panic. Program suspended; original records still in use. Total spend: $14M, value delivered: ~$0.
Investment
$14M
Patient Records Targeted
~3M
Clinician Adoption
<5%
Outcome
Suspended
MDM in regulated/safety-critical contexts requires named domain experts with veto authority and explicit accountability. Without that ownership, no one will use what IT builds.
Decision scenario
The Customer Master Mandate
You're newly hired as Director of Data Governance at a $1.5B B2B distributor. The CEO has given you 18 months and $6M to build a Customer Master that finally lets sales, finance, and marketing use the same customer numbers. You discover the regional sales VPs each maintain their own customer hierarchies and view a central master as a threat to their autonomy.
Budget
$6M / 18 months
Duplication Rate
~28%
Conflicting Customer IDs
~120k of ~850k
Regional VP Buy-In
Low
Executive Sponsor
CEO (verbal)
Decision 1
Your team wants to start by building the technical MDM platform and migrating data; the regional VPs are demanding to see the proposed customer hierarchies before they agree to anything.
Move fast on the technical build — install the MDM tool, ingest data from all systems, produce a draft golden record, and present it to the regional VPs at month 6.Reveal
Spend months 1-4 on political work: 1:1s with each regional VP to understand their hierarchies, co-design survivorship rules, get CEO to formally chair a Customer Master Council with VPs as voting members. Then build the platform in months 5-15 with VPs as co-owners.✓ OptimalReveal
Related concepts
Keep connecting.
The concepts that orbit this one — each one sharpens the others.
Beyond the concept
Turn Master Data Management into a live operating decision.
Use this concept as the framing layer, then move into a diagnostic if it maps directly to a current bottleneck.
Typical response time: 24h · No retainer required
Turn Master Data Management into a live operating decision.
Use Master Data Management as the framing layer, then move into diagnostics or advisory if this maps directly to a current business bottleneck.