HR Open Schema: Portable Candidate Data Without Pipeline Drift

A VP Sales promises a start date to a strategic customer. Two days before the offer, your background check vendor rejects the candidate because the identity verification packet is missing the document type and country fields. Those fields existed last quarter, but your new scheduling tool overwrote the candidate record with a slimmer payload. Nobody saw the drift because the integration still returned 200 OK. This is the failure mode RevOps gets blamed for: speed goes down, cost goes up, and you cannot explain the discrepancy cleanly to Finance, Legal, or the board. By the end, you will know how to implement an HR Open based canonical schema, version it, map every tool to it, and roll it out with canaries and rollback so candidate data stays portable and auditable.

Recommendation: define one canonical schema (HR Open aligned) and treat every vendor payload as an adapter to that schema, not the other way around. This prevents tool swaps from redefining your business data. HR Open provides a shared vocabulary for core HR and recruiting entities (candidate, job, application, stage). You do not need perfect coverage of every vendor field. You need a stable contract for the fields that run your business: identity gating status, interview events, assessment results, offer states, and audit references. For RevOps, portability means you can change tools without breaking: (1) stage conversion rates, (2) time-in-stage metrics, (3) risk controls like Risk-Tiered Verification, and (4) evidence trails that protect your reputation when something goes wrong.

Recommendation: write down a single ownership matrix before you write code. Schema governance fails when everyone assumes someone else is watching drift. Owner model that works in practice: Recruiting Ops is the business owner of stage semantics and required fields. Security owns identity and biometric constraints (what is collected, retained, and who can access). RevOps owns metric definitions, dashboards, and the change control gate for anything that affects reporting. Automated vs reviewed: schema validation is automated on every inbound event; high-risk verification outcomes go to a manual review queue; mapping exceptions create tickets with an SLA so they do not rot. Sources of truth: ATS is the system of record for application and stage progression; verification service is the system of record for identity outcomes; interview and assessment systems are systems of record for their raw evidence, but must publish normalized results back to the canonical schema.

Use this as a starter contract. It is intentionally opinionated for hiring integrity: it makes verification and evidence first-class fields so they cannot disappear into notes.

Avoid these because they increase funnel leakage and reduce defensibility under audit.

IntegrityLens AI is the first hiring pipeline that combines a full Applicant Tracking System with advanced biometric identity verification, AI screening, and technical assessments, so your canonical schema has one primary home instead of four disconnected tools. In practice, teams use IntegrityLens to keep HR Open style portability while also enforcing Risk-Tiered Verification and producing audit-ready Evidence Packs. IntegrityLens supports the full pipeline: Source candidates - Verify identity - Run interviews - Assess - Offer. It is used by TA leaders and recruiting ops to run workflow, by CISOs to approve identity and data controls, and by operators who need clean webhooks and observability for integrations.

The Checkr manager survey (31% reporting a false identity interview) implies identity fraud is not rare enough to ignore, so schema fields that carry verification results should be required, not best-effort. It does not prove your loss rate or the effectiveness of any specific control, because it is self-reported and depends on detection and awareness. The Pindrop pipeline observation (1 in 6 remote applicants showing signs of fraud) implies remote hiring pipelines should assume adversarial behavior and instrument for it, including structured fields for risk tier and evidence references. It does not prove causality for any single vendor or method, and "signs of fraud" is detection-dependent. For RevOps, the operator takeaway is simple: if you cannot trace identity and assessment outcomes through one schema and one candidate_id, your funnel metrics are not decision-grade.