You are the analytics lead for Amazon Clinic’s virtual primary care product, which supports ~2.5M visits/month across asynchronous chat and scheduled video visits in the US. The business is expanding payer partnerships, and leadership has committed to a care quality and safety bar that must be demonstrated to regulators, employer clients, and medical directors.
Over the last 8 weeks, the Chief Medical Officer flags that patient complaints are up 18% and a major employer account is threatening to pause rollout unless you can prove the department is measuring care performance rigorously and improving it. At the same time, Product claims that a new “faster triage” workflow reduced time-to-first-response by 22% and should be considered a win.
Stakeholders disagree on what “performance of care and services” means:
You need to propose a coherent metric framework that (a) can be monitored weekly, (b) ties to patient outcomes and business risk, and (c) produces actionable levers for improvement.
| Source | What it contains | Granularity |
|---|---|---|
visits | visit_id, patient_id, modality (chat/video), state, visit_created_at, visit_closed_at, clinician_id, specialty | per visit |
messages | visit_id, sender_type (patient/clinician/bot), sent_at, message_type | per message |
clinical_coding | visit_id, diagnosis_codes, risk_level, guideline_pathway_id, antibiotic_prescribed, controlled_substance_flag | per visit |
follow_up_events | patient_id, event_type (repeat_visit, ER_claim, urgent_care_claim), event_at, related_visit_id | per event |
patient_surveys | visit_id, csat_1_5, nps_0_10, free_text, survey_sent_at, survey_completed_at | per survey |
safety_reviews | visit_id, adverse_event_flag, severity, reviewer_id, review_outcome | per reviewed visit |
provider_schedule | clinician_id, available_minutes, staffed_minutes, shift_start, shift_end | per shift |
refunds | visit_id, refund_flag, refund_reason, refund_amount | per visit |
Constraints: