Separate caller-ID authorization, origination, and handset presentation during transfer RCA

Caller-ID RCA across SIP transfers must treat four layers as separate: (1) the inbound caller identity, (2) an owned-number policy that authorizes forwarding, (3) the outbound SIP From/CLI actually submitted, and (4) what the terminating carrier or handset displays. An owned DID stored on the outbound leg can be authorization metadata for Diversion; it does not prove that DID was selected as caller ID. Correlate one exact leg across application operation state and the provider CDR before attributing a fallback or rewrite. Limit: even a provider CDR showing the original ANI establishes the provider’s signaling/accounting view, not the final handset presentation; closing that last boundary requires terminating-side evidence or a controlled second destination.

Evidence/example — 2026-09-10

In the negotiation LiveKit/Telnyx transfer investigation associated with COR-3991, two sanitized staging calls had original ANI present and distinct from the receiving DID, exactly one answered nonfallback leg, and no provider failure. Code set normal-leg sip_number to original ANI while using the owned receiving DID in Diversion. Correlated Telnyx SIP-trunking CDRs recorded caller_number, cli, and sip_from_url as the original ANI, with successful D00 and B-level STIR/SHAKEN attestation. A prior interpretation of the persisted owned-DID policy as the chosen ANI was corrected: that policy authorizes the receiving DID. The handset was reported to show the owned DID, so the remaining qualified hypotheses are terminating-carrier/handset handling of redirecting information or mis-correlation with an earlier fallback call. Provenance: repository code at origin/livekit_migration; read-only sanitized staging Supabase rows; read-only Telnyx detail-record API. No raw phone numbers or call identifiers retained.