Operations Coverage & Productivity

Loading…
This weekLast week This monthLast month 7d30d 90dAll
Outreach + missed inboundAll (incl. BCMs)

Today's call plan

Who to call, when — built from open referrals, the 2-week agent capacity pattern, and historical connect rates by age × time of day. Each window is sized to capacity and filled with the highest-connect-probability patients. The named worklist (with patients + assigned window) is generated locally each refresh; this view is the PHI-free plan + live progress.

Connect rate by age × time window (call reference)
Agent capacity (calls/day by window, last 14d)

Hourly coverage — calls by agent & hour of day (ET)

Who is on the phones, and when. Darker = more calls placed/received in that hour. Use this to spot coverage gaps and overlap.

FewerMore calls

Time of day — when to call & when people pick up (ET)

OutboundInboundAll
Connect rate %Call volumeConnected volume

Day of week × hour

Volume & connection over time

Stacked by connection status per day. Hover a bar for counts.

Connected Answered (untranscribed) Voicemail Not connected

Agent productivity

Click a column header to sort. Conn % = confirmed two-way ÷ total dialed. Columns marked * are rates over connected calls: Enroll %* = enrolled ÷ connected, Decline %* = declined ÷ connected. Use the call-type toggle to see volume + outcomes per kind of call.

AllEnrollmentReschedulingInboundInsuranceReminderOther

Agent activity by day

Calls per agent per day — spot who's dialing, when, and what kind. Respects the team/line filters above.

7 days14 days30 days
AllEnrollmentReschedulingInboundInsuranceReminder

Referral call cadence — attempts → enrollment

Per referred patient (referral channel, joined to calls). True outreach attempts only — calls after a patient enrolls (reminders, care-manager) are excluded. Bars = enrollments first won at each attempt; line = cumulative % captured; dashed = connect rate. SLA = up to 7 attempts, every 2–3 days, until enrolled or declined.

Cadence trend by referral month

Depth of follow-up (median attempts per referral) and resulting conversion (enrolled ÷ referred), month over month.

Outcome breakdown

Every call in the filter, by what happened. Hover for share.

Enrollment funnel

Outbound enrollment calls only — from dial to result. Untranscribed = answered but no recording to read.

Call type mix

What kind of work the team is doing.

By line (campaign)

The dialed number — the cleanest campaign signal.

Data notes & limitations — read before acting on these numbers
Update cadence
  • Refreshes hourly, 6 AM–8 PM ET — but only while the source machine is awake. If it sleeps, data holds until it wakes, then catches up. The banner above shows true freshness.
  • "Called today" lags by up to one refresh (~1 hr) and depends on Aircall logging the call — it's near-real-time, not instant.
  • New-call outcome labels currently fall back to rules (~75%) when the LLM API has no credits; a weekly pass restores full quality. Older calls are LLM-labeled (~90–95%).
Data
  • ~34% of calls have no recording/transcript — human vs. voicemail is unknowable for those (shown as a separate "untranscribed" bucket); connect rate is a bounded estimate, not exact.
  • Call history starts 2026-03-11, so attempts before then aren't counted — affects only old (archive-tier) referrals.
  • Calls are matched to patients by phone (~98%); shared/changed numbers can mis-match. Decline detection uses status text + call outcomes and can misclassify a few.
  • Connect-rate-by-age×window is a 90-day historical average (refreshed periodically, not every run). Proactive vs. referral split relies on tags, which may be incomplete.
  • Agent metrics cover calls attributed in Aircall; ~20% of calls are unassigned. Referral-volume has a few days of logging lag near "today."