Identity verification that got patients answering the phone
Patients ignored calls from unknown numbers, blocking telehealth adoption. Dialer showed the doctor's office number — verified and HIPAA-safe.
Timeframe
2013–2017
Focus
Redacted, de-identified outcomes
My role
Early product lead on Doximity Dialer. Owned identity/verification UX, caller-ID mapping rules, safety copy, and fallback workflows.
Relevance to healthcare AI
When a system decides what identity to present, that decision has to be traceable and correctable. This is the same constraint that applies to clinical AI at scale — model outputs need provenance, override paths, and records that survive audit.
Summary
Patients ignored calls from unknown numbers, blocking telehealth adoption. Dialer showed the doctor's office number — verified and HIPAA-safe.
1 Doctor opens Dialer →2 Enters patient phone number →3 System verifies office caller ID →4 Patient sees "Dr. Smith's Office" on phone →5 Patient answers →6 Call connected with verified identity
Detail
Constraints
HIPAA privacy and patient consent requirements.
Carrier spoofing rules and call deliverability risk.
Clinician adoption risks if workflows added extra steps.
Interventions
Verified caller ID and safety copy.
Audit-ready call logs and escalation guidance.
Fallback workflows for missed connections.
Outcomes
Clinics stopped raising pickup rates as a top concern — the identity problem was resolved, not managed.
iOS app reviews climbed from 3.7 to 4.8 stars after mobile product work.
Spoofing reports dropped below the carrier review threshold — the verification system held under real conditions.
Answering stopped being the bottleneck — clinician adoption accelerated once identity was solved.
Artifacts
Documented how caller ID mapping rules interacted with carrier spoofing protections.
Consent copy and verification UX patterns tested across clinic workflows.
Escalation playbook for failed connections and missed-call recovery.
Key takeaways
Identity is a product decision, not just a feature — who the system says you are shapes whether anyone picks up.
Audit-ready logs and escalation playbooks matter as much as the happy path.
Carrier rules, privacy constraints, and clinician adoption shape the product from the start.
Details are anonymized to preserve healthcare data privacy and proprietary corporate metrics. Where exact figures cannot be shared, I describe the operating problem, intervention, and observed directional change.