Analytics evidence lane

Grounded analytics, proof, and agent value

See what changed, why it changed, and what needs action next without turning hiring metrics into vague AI claims.

Operator command surfaces

Stay inside the same queue, proof, and launch loop

Analytics should now read as the evidence lane for the rest of the operator shell instead of a detached report. Use it to decide whether the next move belongs in applications, providers, onboarding, Agent Hub, or launch.

Current lane · Analytics Applications queue Agent delivery proof Launch truth

Workspace Run the daily queue

Overview MANAGE Pipeline PIPELINE Applications APPLICATIONS Interviews INTERVIEWS Analytics Current

Control Prove, recover, and launch

Agent Hub AGENTS Approvals APPROVALS Providers PROVIDERS Onboarding ONBOARDING Mappings MAPPINGS Launch LAUNCH

Total applications7

Avg days to review0

Avg days to interview0.2

Approved to hire1

Onboarding received0

Stale candidates1

Queue healthDEGRADED

Worker runtimeBLOCKED

Blocked jobs16

Why this changed

Hire movement is backed by approved packets

1 candidate packet(s) are already approved to hire.

POSITIVE

Candidate contact has real receipt proof

1 message(s) show delivered status instead of draft-only outreach.

POSITIVE

Onboarding now supports employers outside the HospiEdge stack

1 handoff(s) are routed through external or manual destination modes.

NEUTRAL

Approved agent work is being followed by downstream movement

1 approved action(s) were followed by later workflow progress.

POSITIVE

Some pipeline drag still comes from stale candidates

1 candidate(s) are still sitting in early-stage stale buckets.

WARN

Async recovery work is still affecting operator trust

16 queue job(s) remain blocked or dead-lettered.

WARN

What needs action

Clear pending approvals

3 approval request(s) are still blocking messages, provider actions, or hire decisions.

Open

Review stale early-stage candidates

1 candidate(s) have aged past the early-stage SLA.

Open

Finish candidate communications waiting approval

1 message(s) are drafted but not yet cleared to send.

Open

Recover failed onboarding pushes

1 onboarding handoff(s) failed and need retry or destination correction.

Open

Work the blocked queue lane

16 queue job(s) still need operator recovery.

Open

Review pending agent actions

1 agent action(s) still need a human decision before they can count as trusted follow-through.

Open

Pipeline aging

0 2 days 3

3 5 days 1

6 plus days 3

Source quality

Source	Applications
site_apply	5
quick_apply	1
referral	1

Agent value proof

Total actions2

Approved1

Follow-through100%

Pending review1

Jordan Reyes · General Manager

Interview scheduled: phone_screen

save_note · 2026-03-29T13:20:00+00:00

Communications proof

Total messages3

Delivery logged33.3%

Delivered50%

Awaiting approval1

Interview throughput

Scheduled2

Completed2

Scorecard coverage100%

Avg panel score8.3

Offer and onboarding proof

Offer packets2

Pending approval0

Approved to hire1

Onboarding received0

Onboarding mode	Count
HospiEdge stack	0
External system	0
Manual checklist	1
Receipt-backed completion	0%

Role health

Which roles are moving and which are dragging

Role	Applications	Stale	Interview %	Hire %	Avg response days
General Manager role-level funnel summary	5	0	20%	0%	0
Kitchen Manager role-level funnel summary	1	1	0%	0%	—
Server Trainer role-level funnel summary	1	0	100%	100%	—

Location health

Location	Open jobs	Applications	Interview %
Cedar House River North location-level hiring signal	1	5	20%
Sunset East Nashville location-level hiring signal	1	1	0%
Harbor Table Scottsdale location-level hiring signal	1	1	100%

Queue health

DEGRADED

Worker heartbeat is stale or missing. Treat async work as at-risk until a worker checks in.

Latest heartbeat: async-worker · 2026-03-29T15:18:00+00:00

queued 14

running 0

completed 1

failed 1

dead letter 1

Worker heartbeat

async-worker

Completed analytics snapshot refresh.

2026-03-29T15:18:00+00:00 · healthy

export-worker

Waiting on provider credentials for ziprecruiter channel sync.

2026-03-29T14:41:00+00:00 · degraded

Worker ping command

Use the CLI below from the worker or cron lane so launch readiness sees a real fresh heartbeat instead of a decorative UI flag.

php scripts/ping_worker_heartbeat.php --worker=async-worker --state=healthy --note="Queue worker heartbeat confirmed." --json

Metric definitions

Days to review

From application created time to the first explicit move into review.

Interview throughput

Count of scheduled and completed interviews plus scorecard coverage on completed interviews.

Offer proof

Only hire decision packets and approval states count. There is no hidden offer status.

Agent value

Counts approved actions and how many were followed by downstream workflow progress. It does not claim direct causation.

Onboarding completion

Only submitted handoffs with receipts can move into received. External and manual lanes stay visible as separate modes.

Master AI surfaces

Agent hub

Review signed contract reads, writes, launch rules, and command examples for the agent app.

Open

Command surfaces contract

Read public IDs, backlinks, safe embed summaries, and launch conventions for cross-app operator context.

Open

Launch truth

See queue health, provider truth, and go-live boundaries from the operator side.

Open

Analytics contract view

Signed API surface the agent app should consume for report-grade analytics proof.

Open

Worker runtime

BLOCKED

Worker heartbeat is stale or missing, so async work should be treated as at-risk.

Latest state: healthy

Queue-SLA summary

14 overdue queued · 0 stalled running · 1 dead-letter

Failed attempts: 1

Record a fresh worker heartbeat

Run the heartbeat CLI from the worker or cron lane so launch readiness can prove that async work is still alive.

php scripts/ping_worker_heartbeat.php --worker=async-worker --state=healthy --note="Queue worker heartbeat confirmed." --json

Clear queued jobs past SLA

14 queued jobs are past the pickup SLA. Retry or archive them from analytics before widening rollout claims.

Open

Clear dead-letter queue jobs

1 dead-letter job still need explicit recovery proof.

Open

Cron runbook

Heartbeat check-in

Keep the worker freshness window alive so launch readiness sees a real heartbeat instead of a cosmetic page refresh.

Schedule: */5 * * * *

cd /home/hospneac/hospiedge.com/hospi_jobs && php scripts/ping_worker_heartbeat.php --worker=async-worker --state=healthy --note="Queue worker heartbeat confirmed." --json

Worker runtime proof

Record a machine-readable queue-SLA and worker-health snapshot for operator review.

Schedule: */15 * * * *

cd /home/hospneac/hospiedge.com/hospi_jobs && php scripts/run_worker_runtime_verification.php --json

Launch readiness proof

Re-check blocker counts after worker/queue movement so the launch surface stays grounded.

Schedule: */15 * * * *

cd /home/hospneac/hospiedge.com/hospi_jobs && php scripts/run_launch_readiness.php --json

Queue model

State machine and worker expectations

Queue states

Queued

Waiting for a worker to pick up the job.

Running

A worker has started processing this job.

Completed

The async task finished successfully.

Failed

The job failed but can still be retried.

Dead letter

The job exhausted retries and needs operator recovery.

Worker expectations

Workflow Worker

automation, reminders, and lightweight follow-through tasks

Should pick up queued work within the job SLA window. Cadence: every 5 min · Fresh within 15 min · Cron */5 * * * *

Master AI Sync Worker

backfill syncs and Master AI-triggered async jobs

Should preserve org/location scope and actor refs end to end. Cadence: every 5 min · Fresh within 15 min · Cron */5 * * * *

Onboarding Worker

handoff pushes and downstream receipt tracking

Should never mark a handoff received without a receipt state. Cadence: every 5 min · Fresh within 15 min · Cron */5 * * * *

Analytics Worker

snapshot refreshes, queue visibility, and launch-readiness evidence

Should keep queue health and launch evidence fresh enough for operator review. Cadence: every 15 min · Fresh within 30 min · Cron */15 * * * *

Queue inventory

What async work exists and who owns it

Job type	Group	Owner	Total	Queued	Failed	Dead-letter
onboarding_pushasync workload	onboarding	Onboarding Worker	14	13	0	1
daily_reminder_runasync workload	automation	Workflow Worker	1	0	0	0
channel_syncasync workload	provider_sync	Provider Sync Worker	1	0	1	0
analytics_snapshotasync workload	analytics	Analytics Worker	1	1	0	0

Recovery lane

Blocked jobs and manual recovery

onboarding_push dead_letter

Downstream onboarding system timed out during packet receipt.

Attempts: 3 / 3 · Owner: Onboarding Worker · Origin: hiring_app

Confirm the onboarding handoff packet is still approved and complete.

Retry the queue job and then check the onboarding receipt state.

onboarding_push queued