Source Quality Triage: Prove Which Channels Actually Perform

Your sourcing report shows record applicant volume. Two weeks later, the interview loop is overloaded, time-to-offer breaches your internal SLA, and a post-mortem reveals that a meaningful slice of interviews never should have happened because identity was unverified or interview performance was consistently below bar from one channel. For a CPO, the risk is not just wasted recruiter hours. It is audit liability and cost exposure: you funded channels based on volume, then advanced candidates without a defensible evidence trail connecting source to performance. Replacement costs are commonly estimated at 50-200% of annual salary, role-dependent, so scaling the wrong source mix is a budget leak you can quantify in Finance terms. (SHRM) Fraud turns "source quality" into a security control. 31% of hiring managers report interviewing someone who later turned out to be using a false identity. If you do not segment conversion and performance by source with identity gating, you cannot isolate which sources are creating the exposure. (Checkr)

Most stacks measure what is easy: applicant counts, clicks, and time-in-stage inside the ATS. They do not measure what is defensible: identity-verified conversion, rubric-calibrated interview outcomes, and early retention proxies tied back to the original source. Why the market failed to solve it: legacy tools split your truth across systems. ATS records stages, interview platforms hold rubrics inconsistently, assessment vendors store telemetry elsewhere, and background checks arrive late. Without an immutable event log and unified evidence packs, you cannot answer basic audit questions: If legal asked you to prove who approved this candidate, can you retrieve it? Operationally, the typical failure pattern is predictable: - Sequential checks that slow everything down. - No unified event logs or evidence packs per candidate. - No review-bound SLAs, so exceptions and edge cases sit in inboxes. - No standardized rubric storage, so "quality" becomes subjective and non-reproducible. - Shadow workflows and data silos, where sourcing insights never reach the interview loop.

Recommendation: assign owners per control point before you change metrics. Source quality becomes reliable only when each step has a named owner, an SLA, and a log artifact written back to the ATS. Ownership model (minimum viable): - Recruiting Ops owns workflow design, required fields, queue SLAs, and reconciliation when webhooks fail. - Security owns identity gate policy, step-up verification rules, access control, and audit policy for evidence packs. - Hiring Managers own rubric discipline, structured scoring, and debrief decision logging. - People Analytics owns dashboards, segmentation by source, and monthly channel rationalization.

Recommendation: treat sourcing as an instrumented, risk-tiered funnel where every downstream outcome is attributable to a source and anchored to an identity-verified candidate record. Operating model components:

IntegrityLens AI fits as the ATS-anchored control plane that ties source to downstream, identity-verified outcomes with tamper-resistant logs. The goal is not more data. The goal is defensible, timestamped evidence that survives audit questions and budget reviews. What it enables operationally: - Identity gate before interview access using liveness checks, face match, and document authentication, with typical end-to-end verification in 2-3 minutes. - Risk-tiered verification so higher-risk sources and remote roles step up controls without slowing every candidate. - AI screening interviews available 24/7 to normalize early-stage signal capture across timezones. - Technical assessments across 40+ languages with plagiarism detection and execution telemetry for evidence-based scoring. - Immutable evidence packs: timestamped logs, reviewer notes, and audit-ready exports with zero-retention biometrics options.

Do not do these three things if you want source quality metrics you can defend: - Let scheduling happen before identity verification for remote roles or flagged sources. It creates privileged access without gating and wastes interview capacity. - Store rubrics in free text or attachments that cannot be aggregated, compared, or audited. If it is not structured and timestamped, it is not defensible. - Run exceptions in Slack or email without writing back approvals to the ATS event log. Shadow workflows are integrity liabilities.

Recommendation: implement source quality tracking as a staged control rollout. Start with clean source data, then identity gating, then rubric standardization, then retention linkage. Step-by-step (with SLAs, owners, and evidence):