Leads inactivity manager digest (1.9.0) — Production Readiness Audit

Field	Value
Date	2026-10-05
Status	Go for production (migrate catalog + deploy SPA + restart emails queue)
Scope	Replace per-lead inactivity bursts with one assignee digest and one manager table email; keep assignment; Settings → Leads knobs; catalog leads 1.8.0 → 1.9.0
Companion	Leads developer guide · Tenant settings · Upgrade · CHANGELOG

---

Executive summary

Idle assigned leads no longer fire one in-app/push notification per record. The existing daily leads:notify-inactive job groups idle leads per assignee and per manager (department manager, else workspace owner). Assignees get one bell/push digest. Managers get the same plus one branded HTML email with clickable lead name, phone, and email. Leads stay assigned. Won/Lost stages remain excluded. New workspace settings control Active-only, skip future follow-ups, manager email, assignee notify, and reminder cadence.

Go / No-Go: Go.

Gate	Result
Platform freeze (existing settings + notification stack; no parallel shell)	Pass
Assignment unchanged	Pass — command never writes assigned_to
Won/Lost still skipped	Pass — Pest
Active-only skips Closed	Pass — Pest
Future pending follow-up skipped	Pass — Pest (UtcInstant vs due_at)
One digest per recipient (not per lead)	Pass — Pest two-lead case
Manager mail gated by leads.inactivity_manager_email	Pass — Pest
Threshold 0 disables job	Pass — Pest
Reminder period dedupe (default 7 working days)	Pass — Pest
Settings GET/PUT for new keys	Pass — Pest
Catalog migrate-only 1.9.0 + CatalogSeeder companion	Pass
Queued manager mail CTA baked while tenancy is initialized	Pass — listUrl + per-lead URLs built in the command

Findings and remediations

ID	Severity	Finding	Resolution
L-01	High	Per-lead alerts on a backlog (100s of push/in-app)	Digests per recipient; default reminder period 7 working days
L-02	High	First-crossing modulo would skip already-idle backlog	Include all idleDays >= threshold when reminder > 0; period key on recipient dedupe
L-03	Medium	Queued toMail() calling FrontendUrl::spa after tenancy()->end()	Bake listUrl and row URLs in the command while tenancy is live
L-04	Low	Optional email_notifications defaults off would hide manager mail	Gated by leads.inactivity_manager_email (default on)
L-05	Info	Legacy LeadInactiveNotification / LeadInactiveOwnerNotification unused by the job	Kept for already-queued jobs and historical database rows

What shipped

Backend

• Settings: leads.inactivity_active_only, leads.inactivity_skip_pending_follow_up, leads.inactivity_manager_email, leads.inactivity_assignee_notify, leads.inactivity_reminder_working_days
• LeadInactivityService eligibility + digest rows + period dedupe keys
• NotifyInactiveLeadsCommand grouped sends
• LeadInactiveDigestNotification (lead.inactive.digest)
• LeadInactiveManagerDigestNotification (lead.inactive_escalation.digest + branded table mail)
• Catalog migration 2026_10_04_193718_bump_leads_module_to_1_9_0

Frontend

• Settings → Leads controls (underscore RHF keys)
• Notification registry for digest types (list href /leads)
• Playwright: e2e/tests/settings/tenant-settings.leads-inactivity.spec.ts (one demo login session)

Docs

• User/developer tenant-settings, Leads guides, upgrade, roadmap, changelog, this audit

Test evidence

herd php artisan test --compact tests/Feature/Tenant/Notification/LeadInactiveNotificationTest.php
Tests: 12 passed (37 assertions)

vendor/bin/pint --dirty --format agent — pass.

Playwright npm run test:e2e:leads-inactivity (--project=tenant): 1 passed — one-session validate (91 / −1), save 5 / 14 + Active-only off, reload persists, restore 3 / 7 + Active-only on.

Known limitations / deferred

• Manager table cap is 25 rows plus “and X more”; full set is still in the in-app digest count.
• Reminder 0 only includes leads on the exact threshold-crossing working day (missed scheduler day is not backfilled).
• No live SMTP assertion in CI (Notification::fake()). Staging should confirm one manager HTML mail after leads:notify-inactive.
• Historical in-app rows of type lead.inactive remain in the registry for old notifications.

Upgrade & staging smoke

1. php artisan migrate --force — catalog leads → 1.9.0 (do not db:seed).
2. php artisan queue:restart.
3. Settings → Leads: set invalid 91 days → client validation; save 5 / 14 → reload persists; restore 3 / 7.
4. Create two Active assigned leads older than the threshold, run php artisan leads:notify-inactive. Assignee: one digest. Manager: one email with two clickable rows. Confirm assigned_to unchanged.
5. Move a lead to Won: it must not appear in the next digest.

Rollback

Roll back the catalog migration (down() → 1.8.0) after rolling back application code. Settings rows are additive; unused keys are harmless. Restore the previous command/notification classes to return to per-lead alerts.