PUBLIC BETA · 0.1.0rc2 Source-available · ELv2 · Self-hosted

Manage reference data from a single source

Country codes, cost centres, parameters, mappings between applications, organisational hierarchies and rates. Manage them in one place, with typed schemas, review before publication, versioning and an API that supports historical reporting.

Publicly available for pilot users and testers. Not yet recommended for production use.

$ pipx install --pre refrncr
$ refrncr start
→ refrncr running on http://localhost:8000 (SQLite)

The --pre option is required during the public beta. Upgrade compatibility with stable 0.1.0 is not guaranteed.

# Managed refrncr, provisioned per tenant
$ curl https://api.refrncr.com/api/v1/datasets/countries/records \
    -H "Authorization: Bearer $REFRNCR_KEY"

What counts as reference data

  • Cost centres
  • Country codes
  • Currencies
  • Error codes
  • Product categories
  • Shipping rates
  • Chemical elements
  • Parameters
  • GL accounts

Why spreadsheets stop working

Spreadsheets work—until multiple teams depend on the same data.

Spreadsheets are useful tools, but they are not a reliable system of record for shared reference data. Files such as country_codes_v3_final_FINAL.xlsx lack an enforced schema, validation and version history, making it difficult to reconstruct the data used for a previous reporting period.

Today

  • country_codes_v3_FINAL.xlsx SharePoint
  • countries_2026.csv Ops laptop
  • dim_country warehouse, hand-edited
  • COUNTRY_LOOKUP = {…} 4 repositories
  • Countries (Sheet2) the finance model

No owner · no version · no history

Turning point

refrncr

Dataset
countries
Unique key
iso_code_3
Owner
Data Management
Version
Published v12
Effective
2026-01-01 →

Reviewed, versioned, auditable

Every consumer

$ curl \
    -H "Authorization: Bearer $REFRNCR_KEY" \
    "$REF/api/v1/datasets/countries/records"
{
  "data": [
    { "identifier": "…",
      "values": {
        "iso_code_3": "ABW", … } },
    …
  ],
  "pagination": { "total": 250, … },
  "version": {
    "number": 12,
    "effective_from": "2026-01-01"
  }
}

One source, same answer everywhere

Three shapes of reference data

Reference data is more than a code list.

Some reference data is a list of values. Other data defines relationships between systems or reporting structures. Refrncr supports all three using the same review, versioning and publication process.

Datasets

Typed, governed lists with stable business keys, validation rules and immutable published versions.

Examples · shipping rates · countries · cost centres

Explore reference datasets

Mappings

Crosswalks that relate codes between systems and immediately expose missing or invalid mappings.

Example · SAP cost centre → Salesforce territory

Explore mapping datasets

Hierarchies

Hierarchies that aggregate or provide insight into a dataset, with parent–child relationships and effective dates.

Examples · organisation structure · regional classification

Explore hierarchy datasets

Explore reference data management

The product

From draft to publication.

Reference data should not be edited directly in production. Work in a draft, review changes and publish a new version when it is ready.

A typed dataset, not a sheet

  • Create a custom schema with real types: string, number, boolean and date.
  • Define required and unique constraints on a stable business key.
  • Create and edit records in an explicit draft, never directly in a published version.
  • Import CSV and Excel files, map the columns and review the proposal before confirming the import.
The Fields tab of the Cost centres dataset in refrncr, in a draft: Code is the identifier and is required and unique; Name, Start date and Reportable are required. Each field has a fixed data type: string, date, boolean or decimal.

Point-in-time API

Report with the dataset that applied at the time.

Reference data changes over time. Historical reports should use the data and structure that were effective during the reporting period—not whatever is current when the report is generated.

dataset · effective_date
# 1 · no parameters → current published version
$ curl -H "Authorization: Bearer $REFRNCR_KEY" \
    "$REF/api/v1/datasets/cost_centers/records"

# 2 · January–February → version 15 (Sales)
$ curl -G -H "Authorization: Bearer $REFRNCR_KEY" \
    "$REF/api/v1/datasets/cost_centers/records" \
    -d effective_date=2026-02-28

# 3 · version 15, checked against that date
$ curl -G -H "Authorization: Bearer $REFRNCR_KEY" \
    "$REF/api/v1/datasets/cost_centers/records" \
    -d version=15 -d effective_date=2026-02-28

# 4 · March–June → version 16 (Operations)
$ curl -G -H "Authorization: Bearer $REFRNCR_KEY" \
    "$REF/api/v1/datasets/cost_centers/records" \
    -d effective_date=2026-03-01
DATASET VERSIONS V15 Cost centre → Sales effective Jan–Feb V16 Cost centre → Operations effective from 1 Mar 1 Mar REPORT 15 Jul each date resolves to its effective dataset version

A cost centre moves from Sales to Operations on 1 March. A report for the first half of the year, generated on 15 July, attributes it to Sales for January and February and to Operations from March through June. Refrncr preserves both states, so the report remains historically accurate and reproducible.

What you stop maintaining

One source for shared reference data.

Today

Duplicate source files

With refrncr

One governed dataset

Maintain each dataset in one place, with a named owner, a defined schema and validation rules. Systems consume the published data instead of keeping separate copies.

Today

Version names in filenames

With refrncr

Fixed, published versions

Edit in a draft and publish an immutable version when it is ready. The version history shows exactly which data was available at any point in time.

Today

Mappings embedded in code

With refrncr

Mappings as governed data

Manage mappings between systems as governed data. Every consumer uses the same published mapping, while missing and invalid mappings remain visible.

Today

Incomplete change records

With refrncr

A complete audit trail

Keep the author, review decision, effective date and field-level changes together. The audit trail is created during the change process, not reconstructed later.

Editions

The same product, self-hosted or managed by us.

Start with the self-hosted edition and move to a managed deployment when you no longer want to operate the platform yourself.

Source-available · ELv2

Community

Self-hosted, single organisation per installation. Start on embedded SQLite with zero infrastructure; move to PostgreSQL when it becomes a real production system.

Managed SaaS

refrncr Cloud

Coming soon

Custom

Enterprise

Coming soon

CapabilityCommunityCloud Coming soonEnterprise Coming soon
Datasets, typed schema & records✓
Mappings & crosswalks✓
Hierarchies✓
Draft → review → publish✓
Effective dating & point-in-time API✓
Teams, permissions & audit✓
Multi-user, multi-team✓
Embedded SQLite quick start✓
PostgreSQLbring your own
Multi-tenant platform—
Provisioning, quotas & billing—
SSO / SAML / OIDC, SCIM—
EU data residencyyour infrastructure
SLA & commercial support—

✓ included · optional = available by agreement · — not applicable to this edition · planned = not shipped yet

Pricing

Community is free. Cloud and Enterprise are coming later.

Community is free, and stays free. Cloud and Enterprise are coming soon.

Self-hosted · ELv2

Community

€0forever

Unlimited editors · your infrastructure

  • The complete core: datasets, mappings, hierarchies, versioning, API
  • Multi-user and multi-team, one organisation per installation
  • Embedded SQLite quick start, or your own PostgreSQL
  • Permissions and audit included
  • No licence or usage fees
  • Community support
Managed SaaS

refrncr Cloud

Coming soon

Custom

Enterprise

Coming soon

Non-goals

What refrncr is not.

Refrncr is designed for governed reference data that changes deliberately and relatively infrequently — not transactional or operational data that changes throughout the day. These are deliberate product boundaries, not gaps waiting to be filled.

  • Golden records and survivorship
  • Calculated measures and allocations
  • Deduplication and fuzzy matching
  • Customer MDM or PIM
  • A data catalog or lineage product
  • A generic data-quality engine
  • A product catalogue
  • A BI platform
  • A generic BPM or workflow engine

Running it

SQLite to start. PostgreSQL when it matters.

Quick start

Embedded SQLite, no infrastructure. One install, one command, a working instance with your own data in it.

Bring your own DB

PostgreSQL for serious multi-user production use, in your own network, under your own backup and retention rules.

Permissions

Effective permissions in business language — View, Edit, Approve, Publish — with team-derived and direct grants both inspectable.

API keys

Secrets are shown once at creation. Afterwards you see the identifier and metadata; key creation and revocation are audited events.

Deploy

refrncr.org

Install refrncr and try it with your own reference data.

Source-available under ELv2, self-hosted and free to run forever. Import one real list — your cost centres, not a sample file — and publish it through the API before you decide anything else.

$ pipx install --pre refrncr
$ refrncr start
refrncr.com

Prefer a managed deployment?

We can operate the same platform for you, including hosting, backups and platform maintenance.

  • Managed PostgreSQL, no infrastructure to staff.
  • Multi-tenant, still multi-user and multi-team.
  • Enterprise capabilities on request as they ship.