What is the recommended structure for a Siberson Veriket Data Classification Proof of Concept?
| POC Element | Recommended Parameters |
|---|---|
| Duration | 2–4 weeks |
| User scope | 25–100 users across 2–3 representative departments — recommended mix: Finance (PCI/financial data), HR (PII-heavy), and one business unit with significant document production volume |
| Classification taxonomy | Deploy a simplified 3–4 level taxonomy for POC purposes (e.g., Public, Internal, Confidential, Restricted) — validated against the prospect's existing classification policy if one exists |
| Policies to activate | 3–5 targeted automated classification policies covering the prospect's highest-priority data categories — typically: PII detection, financial data patterns, and a custom keyword-based policy for proprietary content |
| OS and application coverage | Deploy across the full range of OS and productivity application combinations in the target environment — particularly important if Linux, Pardus, or LibreOffice is in scope |
| DLP integration test | If the prospect has an existing DLP platform, configure it to read Veriket classification labels during the POC — demonstrating the immediate DLP accuracy improvement from classification-driven enforcement |
| Success metrics (agreed upfront) | Classification coverage rate (% of documents labeled), automated classification accuracy rate (% correctly labeled without user correction), false positive rate, time-to-classification per document, user adoption rate, audit trail completeness |
| Exit deliverable | POC summary report: classification distribution across the POC population, automated vs. manual classification ratio, accuracy assessment, DLP integration test results, user feedback summary, and recommended production taxonomy and policy configuration |
Last updated: 2026-04-12