Reltio

Overview

Reltio is a cloud-native multidomain master data platform, now positioned around context intelligence rather than master data management, with its own agent framework and identity builder alongside the core.

Ownership changed recently and older descriptions are wrong: SAP completed its acquisition in May 2026, making it an SAP company, though Reltio says the portfolio remains available standalone for the foreseeable future.

Implementation is model-driven rather than a packaged application with a fixed model. You define entity types, attributes and relationships, with an ontology builder that translates existing schemas into its canonical model.

It is multidomain out of the box rather than sold per domain, with industry accelerators shipping canonical models, match rules and survivorship for life sciences, financial services, insurance, healthcare, supplier and product.

The most substantive architectural fact, and the one most often got wrong, is how its golden record works. Records are physically merged into one profile, but each source's values persist as crosswalks.

The golden value is then calculated just in time whenever the entity is requested, so it is a computed view rather than a stored row. Change a survivorship rule and it takes effect immediately with no reprocessing.

Matching supports four group types, from boolean rule formulas through to relevance-based arithmetic scoring, using configurable comparators such as Jaro and Levenshtein distance.

It leans hardest of the four on language models for entity resolution, with an AI matching engine pretrained for the task, marketed as rule-free matching and sold as an add-on.

Its identity builder is distinctive, matching consumers against decades of authoritative referential data, which none of its rivals publishes.

Deployment is the most restrictive here: software as a service only, with no self-hosted option at all, across eleven published regions on three clouds.

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

• Multidomain out of the box rather than sold per domain, covering parties, products, locations and organisations
• The data model is definable, built with an ontology builder and a console data modeller
• Industry accelerators ship canonical models for life sciences, financial services, insurance and healthcare
• Supplier and product accelerators add preconfigured interfaces, match rules and integrations
• Each accelerator includes preconfigured cleansing, matching and survivorship rules

Matching and survivorship

• Four match group types, automatic, suspect, custom and relevance-based with arithmetic scoring
• Rules combine token classes, comparator classes and cleansers, including Jaro and Levenshtein distance
• An AI matching engine pretrained for entity resolution is sold as an add-on, marketed as rule-free
• Survivorship is configured per attribute, with strategies including last update, source priority and frequency
• The golden record is computed just in time over retained crosswalks, so rule changes apply without reprocessing

Stewardship and workflow

• Stewards work in a hub with a potential matches view showing candidates side by side
• A workflow engine assigns review tasks by role using a least-used assignment policy, worked from an inbox
• Unmerge is supported, which is why rule-based matching is recommended over the match-before-create API
• Automatic unmerge is configurable and can optionally include manual merges
• A security audit log exports records in CSV or JSON, filterable by address, user, event type and time

Hierarchies, quality and governance

• Automated householding and grouping of entities for hierarchy management
• Reference data management as a published module, shipped inside the industry accelerators
• Data quality monitoring with a dashboard and AI-assisted insights
• Address validation and geocoding, and third-party enrichment, both as add-on subscriptions
• Integration with a data catalogue is referenced generically without naming specific products

How mastered data is consumed

• REST and GraphQL APIs with a public developer portal
• Streaming change events out of the tenant through external queue configurations
• A data delivery network published as serving under fifty milliseconds
• An MCP server exposing trusted data to AI agents, with the hosted version carrying a platform fee
• Reverse integration to sources runs through integration hub flows rather than a separate feature

Integrations

• More than a thousand pre-built connectors and integration workflows claimed
• An integration hub metered in tasks, either included in the base package or sold as an add-on
• Published partnerships with Databricks, Google Cloud and Microsoft Azure Marketplace
• SAP is now the parent as well as an integration point
• Individual connector names are not enumerated on the pages read

Where it runs and what it costs

• Software as a service only; no self-hosted or private-network option is published
• Eleven published regions across AWS, Google Cloud and Azure, including China and the UK
• Database nodes spread across at least three zones, replicated and active across all of them
• Production, test and development tenants, with the named test tenant matching production volume
• Priced on consolidated profiles, with storage and integration tasks metered separately

Pricing

Price on requestQuote only; metered on consolidated profiles

Quote-only, with no rate card or published amount. The primary unit is the consolidated profile, a count of mastered profiles specified per tenant in your licence, with storage metered separately in its own units and the integration hub metered in tasks with overage per million. Development tenants are entitled to 200,000 profiles. Named add-ons, all quoted, include the AI matching engine, encryption key management, private networking, an immutable vault, address validation and tenant cloning. A free identity tier was announced in 2021 but is not confirmed current.

Vendor pricing page →

Demos and videos

• Reltio platform overview Product demo · YouTube
• Entity resolution and governance for data stewards Tutorial · YouTube

About Reltio, an SAP company

Reltio was founded in 2011 by Manish Sood, who had previously worked on the first master data platform at Informatica, and launched what it calls the first cloud-native platform in the category in 2012. SAP announced its acquisition in March 2026 and Reltio confirmed completion that May, on undisclosed terms, so it is now a wholly-owned SAP subsidiary. Before that it had raised $237m, most recently $120m in November 2021 led by Brighton Park Capital at around a $1.7bn valuation. It reported $185m in annual recurring revenue and more than 200 large enterprise customers for the year to January 2026.

Founded 2011 · Redwood City, California · reltio.com