Test coverage against a working question
Select a supplier, project or dependency your team already needs to understand. Ask whether the dataset identifies the relevant operation, product stage and source evidence. Discovery breadth matters, but a list of names is not equivalent to production history or documented commercial relationships. Define the research output before comparing library sizes.
The USGS occurrence database is a clear example of a source with a specific geological purpose. Enriching such records into industrial research requires additional work; it is not achieved by relabelling every entry as a facility.
USGS: Global rare earth element occurrence database (2022) ↗
Inspect the difficult cases
Test aliases, overlapping source releases, company-level agreements without named sites and observations with different units. A useful dataset preserves these distinctions rather than hiding them behind a neat interface. Check how unknowns and conflicting disclosures are presented and whether you can inspect the original source.
Then test the workflow. Can the team compare the relevant evidence, retain investigation inputs and revisit the question after updates? The value of maintained research is not just retrieval: it is reducing repeated reconciliation while keeping the decision basis understandable.
Illustrative evaluation task
Ask the product to support a shortlist of two suppliers and explain whether they share an upstream dependency. Count the unsupported assumptions required to reach the answer. A smaller, well-evidenced route may be more useful for that decision than a larger directory that leaves the analyst to reconstruct every relationship.
Evaluate usability of the evidence
| Dimension | Test |
|---|---|
| Identity | Can aliases and site scope be resolved? |
| Metrics | Are product, unit and period retained? |
| Relationships | Is the edge meaning and source inspectable? |
Your review checklist
- Use a live research question in the evaluation.
- Distinguish occurrences, projects and operations.
- Inspect unknowns and conflicting evidence.
- Test repeat review as well as initial discovery.
Investigate with Oreline
Oreline combines its curated dataset with evidence, comparison, network and scenario tools for this work. Start with the public sandbox to inspect an investigation, then use the organisation workspace for your team's continuing research priorities.
Follow the evidence through the supply chain.
Explore Oreline’s guided investigations. No account required.
Explore the sandbox →Frequently asked questions
Is the largest dataset necessarily the most useful?
No. Relevance, evidence quality and a usable workflow determine whether it answers the question. Counts should be interpreted by record type and scope.
Does curated mean exhaustive?
No. Curation concerns how research is organised and reviewed. Coverage limits and unresolved information should still be visible.