Bedrock Claim Science

Trust

Data Practices

Effective

Note: these practices are provided for launch readiness and should be reviewed by counsel before the public launch.

Separation of research from operations

Claim Science does not store claim photos, policies, estimates, payment records or sensitive consumer claim uploads. Those belong in the secured Bedrock Claim Score application environment. Research and model development work from governed extracts. Never from raw operational files.

Anonymization and aggregation

Before any claim data is used in research or published findings, it is anonymized and aggregated:

  • Direct identifiers (names, addresses, policy numbers, claim numbers) are removed.
  • Findings are published only at aggregate levels where individuals cannot reasonably be re-identified.
  • Small samples that could identify a property or household are suppressed or combined.

Publication governance

  • Minimum sample sizes are defined before publishing carrier- or region-specific findings.
  • Methodology notes accompany every quantitative report: what was measured, how, over what period, and with what limitations.
  • No silent updates: published findings are versioned or corrected transparently; historical results are never quietly rewritten.
  • Comparative or outcome-based claims, including any carrier comparisons or represented-vs-unrepresented analyses, require legal and compliance review before publication. We do not publish carrier “rankings” or outcome comparisons until sample design, definitions and review are strong enough to defend them.
  • Operational models vs. marketing: internal prediction models are kept separate from public claims. A number the site publishes is a number the methodology can defend.

Material relationships

When Bedrock Claim Science research discusses or promotes a Bedrock service, the relationship is disclosed. Research is presented as the work of the Bedrock ecosystem, not as independent third-party validation.

Data subject requests

Because research datasets are anonymized and aggregated, we generally cannot identify or extract an individual’s data from published research. For questions about personal information provided directly to us (for example, through inquiry forms), see our privacy policy or contact us.

Questions

Data-practice questions: contact Bedrock Claim Science.