Custom datasets
Custom datasets built around your requirement.
Bring Rillor a difficult data domain or monitoring requirement. Rillor can define the source universe, preserve the evidence, structure and label the records, document the dataset, and scope an appropriate delivery method.
When custom data is the right call
No vendor covers the domain
Existing products miss the geography, field set, entity type, or cadence you need.
Analysts maintain the pipeline
Brittle spreadsheets and one-off scrapers are absorbing skilled time.
An AI system needs documented data
Model inputs must be reproducible, traceable, and explainable to reviewers.
Provenance is being lost
Collection is destroying origin context and source rights information.
Change must be tracked
You need to know what moved, when, and against which prior version.
Someone must keep it running
The dataset needs an owner after the first build, not just a delivery.
Requirement variables
Seven variables describe most requirements. Partial answers are enough to begin — completing them is what the first engagement stage is for.
- sources
- entities
- fields
- geography
- history
- cadence
- delivery
Delivery is scoped per engagement: an export, an agreed structured format, or an API scoped to the delivered dataset. Rillor does not publish a general public data API, and access to any dataset is confirmed per request.
Engagement stages
Each stage produces something inspectable. Feasibility and rights are assessed before build, not after.
- 01
Define
Requirement, entities, fields, geography, history, cadence, delivery.
- 02
Acquire
Source feasibility, rights assessment, retained material.
- 03
Structure
Schema, labels, entity resolution, normalization rules.
- 04
Verify
Quality review, reconciliation, documented limitations.
- 05
Deliver
Scoped API, export, or agreed structured format.
- 06
Maintain
Collection cycles, change detection, versions, support.
What you receive
A custom engagement delivers a documented dataset asset — not a one-time extract with no explanation attached.
Feasibility boundary: sources, rights, geography, timeline, and price are assessed case by case. Rillor will say when a requirement is not feasible as specified.
Schema
Declared fields, types, labels, and relationships.
Records
Structured records with resolved identities.
Provenance
Source reference, capture context, rights posture.
Lineage
What changed between source and dataset record.
Versions
Dataset versions with change histories.
Quality notes
Review steps, reconciliation decisions, open issues.
Limitations
What the dataset does not support.
Agreed delivery
The scoped method, cadence, and support boundary.
Shapes of work, drawn from current domains
Custom work is not limited to these domains
Authority material
A bounded corpus with authority identity, document structure, and source-linked relationships.
Hardware market
Specifications and separately classified market observations across a defined configuration set.
Procurement
Awards, suppliers, and items with identifier resolution across inconsistent publishers.
Event monitoring
Communications and classifications aligned to a time base for reproducible event study.
Ready to scope it?
Bring whatever you have — sources, entities, fields, geography, history, cadence, delivery target. Partial requirements are fine; the first stage is defining them.