Informatica MDM

Overview

Informatica's master data line is sold as MDM and 360 applications on its wider data management cloud, and its implementation style is a genuine hybrid rather than one or the other.

It is the only one of the four that ships named pre-built domain applications, for customer, product, supplier, finance and reference data, sitting on a business entity model you can extend.

That model is definable underneath: a tree of nodes mapping to base object tables, built in a provisioning tool, with domains decoupled from implementation so a customer domain can become a person, patient or policy holder.

Its pricing is the commercial opposite of Profisee's. Master data is metered on a per-domain records basis, so adding a domain adds cost, which matters for any multi-domain programme.

Ownership changed and no older description survives it: Salesforce completed its acquisition in November 2025 at $25 a share, valuing Informatica at around $8bn net of its existing stake, so it is no longer publicly traded.

For a vendor-neutral directory that carries a real consideration, because the master data roadmap now runs through a customer relationship vendor's agent platform.

Matching is genuinely probabilistic, with fuzzy match base objects considering spelling variations and misspellings, organised into rule sets of match columns, with published guidance to evaluate fuzzy rules before exact ones.

The golden record is physical and persisted. Nodes map to base object tables in a hub backed by a reference store, and the on-premises base object style is merge style, selected by default.

Unusually among the four it retains a long-lived self-hosted line, with the 10.x on-premises product still documented, so it is not a cloud-only choice.

Its 2026 direction is agentic and headless, with what it calls the first agentic multidomain master data management, a steward agent automating match and quality resolution, and delivery through APIs and agent protocol servers.

Key features and capabilities

The same headings are used for every master data management entry, so two tools can be read side by side.

Domains and data model

• Pre-built applications for customer, product, supplier, finance and reference data, plus a multidomain edition
• The only one of the four shipping named packaged domain applications rather than only accelerators
• Underneath, the business entity model is definable as a tree of nodes mapping to base object tables
• Business entities decouple domain from implementation, so customer can become person, patient or policy holder
• Multidomain is supported but metered per domain, which affects the cost of adding one

Matching and survivorship

• Both fuzzy and exact matching, with fuzzy described as probabilistic and handling spelling variations
• Match rule sets are built from match columns examined for points of similarity
• Published guidance evaluates fuzzy rules first then exact, balancing candidate recall against noise
• Survivorship consolidates on best-value rules with ranking, and identifier survivorship is itself rule-governed
• The golden record is physical, persisted in base object tables, with merge style selected by default

Stewardship and workflow

• A steward framework on-premises, and a console plus the application interfaces in the cloud
• Configurable role-based views of dashboards, hierarchies and network graphs
• A steward agent introduced in 2026 automates resolving quality issues and matching records
• Master data is positioned as carrying full lineage and transparency for autonomous decisions
• Merge, unmerge, task queue and approval specifics were not reachable on first-party pages

Hierarchies, quality and governance

• Hierarchy management through role-based hierarchy and network graph views
• Reference data management as its own application, covering sets, hierarchies and cross-walks
• Data quality and validation are separate services on the same platform, metered in the same units
• Governance catalogue integration is explicit, scanning master and reference data to capture metadata and lineage
• Its conversational AI layer is available to eligible master data users through January 2027

How mastered data is consumed

• Headless delivery through APIs and agent protocol servers is the flagship consumption story
• Business entity services are the documented API layer for reading and writing entities
• Golden record publishing to Databricks, with connectivity and catalogue tag extraction
• Its AI layer operates headlessly, delivering master data inside Salesforce, AWS, Microsoft, Databricks or Snowflake
• The technology is being embedded into Salesforce's agent platform following the acquisition

Integrations

• Salesforce is now both parent and integration target, with master data embedded into its agent platform
• Cloud platforms including AWS, Microsoft, Databricks, Snowflake and Oracle Cloud Infrastructure
• A deepened Microsoft collaboration announced in May 2026 for agentic AI and analytics
• The product application interoperates with the customer application as a product information system
• Business entity service APIs plus the wider platform's published data services

Where it runs and what it costs

• Software as a service on its data management cloud is the current default, deployed to a pod
• A long-lived self-hosted line remains documented as the 10.x on-premises product with its own hub
• Clouds referenced include AWS, Azure, Google Cloud and Oracle Cloud Infrastructure
• A published region list for the cloud master data service was not found
• Metered in processing units, with master data specifically on a per-domain records basis

Pricing

Usage-basedQuote only; per-domain records in processing units

Quote-only, with the model published in detail but no amounts. The metering currency is the processing unit, pre-paid annually across the whole platform, and master data specifically meters on a per-domain records basis, making it the only one of the four charging per domain, the direct opposite of Profisee. Volume tiers apply to master data and reduce the cost per unit as consumption rises, though no tier boundaries or rates are published. Its AI layer is included at no extra cost. No free tier or trial length is published.

Vendor pricing page →

Demos and videos

• Informatica MDM Customer 360 overview Product demo · YouTube
• Supplier 360 portal walkthrough Tutorial · YouTube

About Informatica, part of Salesforce

Informatica was founded in 1993 and is based in Redwood City, though its founders are no longer named on any current first-party page. Salesforce signed an agreement to acquire it in May 2025 at $25.00 a share, valuing it at roughly $8bn in equity net of its existing investment, and completed the deal on 18 November 2025, ahead of guidance; it previously traded as NYSE: INFA. Its technology is being embedded into Salesforce's agent platform while it continues supporting its partner ecosystem. It reports more than 5,000 enterprise customers including over 80 of the Fortune 100, and includes its AI layer at no extra cost.

Founded 1993 · Redwood City, California · informatica.com