Reference data management
Reference Datasets
Manage code lists, parameters and other shared reference data with ownership, versioning and controlled publication.
Country codes, cost centres, status values, parameters and business settings often start in spreadsheets, databases or application configuration.
As more systems depend on them, each dataset needs a clear owner, controlled changes, published versions and effective dates.
Refrncr helps teams manage shared reference datasets through review workflows, versioning and controlled publication.
Definition
What is a reference dataset?
A reference dataset contains shared business values used across applications, reports and business processes.
Some datasets define a list of valid values. Others define parameters, settings or business rules.
Examples include Delivery Methods, Error Codes, Product Categories, Cost Centres, Business Settings, Risk Categories, ML Model Parameters, Approval Thresholds and Retention Policies.
Error Codes
AUTH_401SYNC_503API_TIMEOUTDelivery Methods
EMAILPORTALLETTERRisk Categories
LOWMEDIUMHIGHML Model Parameters
temperature 0.7max_tokens 1500top_p 0.9Approval Thresholds
purchase 5000expense 500refund 1000Retention Policies
invoices 7ycontracts 10yaudit_logs 90dWhere it breaks
Why datasets become difficult to manage
A dataset is easy to maintain when it has one owner and one consumer.
The challenge begins when multiple systems rely on the same values.
- Multiple copies
- Different definitions
- Unknown ownership
- No version history
- Different answers across systems
Many copies
shipping_rates_2026.xlsxSHIPPING_LOOKUPshipping_config.jsonRates (Sheet2)Two common patterns
Code lists and parameter datasets
Some reference datasets define a controlled list of values.
Others define parameters, thresholds and settings that systems rely on.
Both patterns use the same ownership, review, versioning and publication process.
EMAILPORTALLETTERVAT_RATE21RETENTION_DAYS365MAX_DISCOUNT10Ownership
Treat datasets as managed products
Each shared dataset needs a stable identifier, clear ownership, an agreed review process and known consumers.
Published versions remain unchanged while editors prepare the next release in a draft.
This makes the dataset a dependable product rather than a by-product of one application.
| Code | Destination | Rate |
|---|---|---|
| NL | Netherlands | 4.95 |
| BE | Belgium | 6.95 |
| DE | Germany | 7.95 |
Controlled change
Manage changes through a controlled process
Changes are prepared away from the current publication, reviewed by the responsible people and released as an immutable version.
Effective dating
Publish versions with effective dates
Changes do not need to become active immediately.
A version can be prepared, reviewed and published before its effective date.
Consumers can continue using the current publication until the new version becomes applicable.
Versions and audit
Know what changed and when
Every published version remains available.
Publication status, effective date, reviewer and change description remain together so teams can explain which values were used.
Product domains
Reference datasets, mappings and hierarchies
Reference datasets store values. Mappings translate values between systems. Hierarchies define parent-child relationships.
Reference Dataset
AUTH_401Authentication FailedSYNC_503Synchronisation ErrorMapping Dataset
CC-4200TERR-EMEACC-4300TERR-BENELUXHierarchy Dataset
- Europe
- Benelux
- Netherlands
- Belgium
- DACH
- Benelux
Examples
Common reference dataset examples
Cost Centres
Shared financial reference data.
Error Codes
Shared application and integration classifications.
Status Values
Shared workflow and application states.
ML Model Parameters
Versioned AI and machine-learning configuration settings.
Business Settings
Application-wide configuration values.
Retention Policies
Compliance-related parameters.
Why refrncr
Why teams use refrncr for reference datasets
- Single source for shared reference data
- Ownership
- Review workflows
- Version history
- Effective dates
- Audit trail
- API access
- Self-hosted
Start with real data
Start with one shared dataset
Choose a dataset already used by multiple systems, such as country codes, shipping rates or business settings.
Replace existing copies with a single published dataset and make it available through the API.