Case Study
AFKOps Social Discovery
Separate public-profile validation workflow for reviewable leads. Matches public social evidence, records confidence and source links, and stops before outreach.
Social discovery map
Public profiles are validated in a separate lane
Selected lead
Public profile search
Evidence + confidence
Human review
No outreach
System Summary
Built a separate social-discovery workflow that validates public profiles for reviewable leads before any outreach decision.
Business Problem
A valid business lead can still have the wrong or missing social account. Blind profile matching creates false positives, wasted outreach, and privacy risk.
What Czar Studios Built
Campaign-scoped Job Orders, public-profile search probes, evidence capture, confidence scoring, deduplication, and a review queue for LinkedIn, Instagram, Facebook, and other public business profiles.
Architecture and Data Flow
A selected lead enters a separate discovery lane. Public search results and visible profile evidence are normalized into candidate matches with source URLs, account and business cues, confidence, and missing-evidence states. A reviewer accepts, rejects, or requests more evidence.
Integrations
Hermes, Supabase, the AFKOps review dashboard, and public web and social-profile evidence visible without login.
Permissions, Guardrails, and Human Review
Public evidence only. The workflow does not access private profiles, send connection requests, message prospects, or approve outreach. Weak matches stay manual_verify or needs_more_evidence.
Testing and Verification
Verified source URLs, account-name and business cues, campaign-scoped deduplication, zero-match handling, and reviewer outcomes through controlled public-profile smokes.
Practical Outcome
Reviewers receive a separate, evidence-backed profile-validation result instead of treating lead research as proof that a social account is correct.
Privacy-Safe Proof Notes
Architecture and review-flow diagrams only; no private accounts, prospect contact details, client records, or session credentials are exposed.