Skip to main content
Nerve Centre KPIs · Audit Profile · Sentiment Settings Crisp is the merchant’s live-chat inbox AND the place Vortex IQ writes audit findings back to (tagged vortexiq:finding). This audit answers four questions: (1) is the auth/token healthy and pointed at the right website, (2) are Vortex IQ findings actually being resolved (resolution rate) or abandoned (>14d no movement), (3) is the backlog under control - by priority, status, blocked, oldest-open, unassigned, and operator overload - and (4) when findings stall, is the unfixed problem tied to a commerce sibling losing revenue (the abandoned-finding → revenue-dip link)?

​
What this audit checks

​
Authentication & access

  • Plugin token (identifier:key) valid - auth on /website//operators/list returns 200
  • X-Crisp-Tier header correct (plugin vs user) for the supplied token
  • website_id resolves to a real Crisp website the token can read
  • Token scopes cover conversations read + segments write (for finding tagging)

​
Findings follow-through

  • VortexIQ findings open (pending/unresolved, segment vortexiq:finding) above 20 - backlog turning into a graveyard
  • Finding resolution rate (90d) below 50% - audit pipeline creating more than the team absorbs
  • Abandoned findings (>14d no activity) above 5 (warn) / 15 (critical)
  • Avg time-to-fix above 14 days (sustained)
  • Bot-only resolutions inflating resolution rate (resolved with zero human messages)

​
Backlog & capacity

  • Oldest open conversation older than 90 days
  • Unassigned open conversations above 5 (warn) / 20 (critical)
  • An operator carrying more than 10 open conversations (overloaded)
  • Blocked conversations above 3 (warn) / 8 (critical)
  • Highest+High priority more than 50% of the open backlog (queue top-heavy)

​
Flow & throughput

  • Weekly resolved count dropped more than 25% versus the prior week
  • Avg cycle time degraded more than 25% week-over-week
  • Avg lead time above 14 days (queue wait dominating)
  • Sprint velocity dropped more than 20% versus prior period
  • Mid-sprint scope add above 15% (committed follow-up being preempted)

​
Cross-channel: revenue-at-risk (the killer area)

  • Abandoned Vortex IQ findings (>14d) whose audit_finding maps to a live commerce/ads/payment regression - the unfixed problem is still costing money (sibling = commerce/ads/payment connector via vortexiqFindingId)
  • Finding resolution rate below 50% sustained while a commerce sibling shows a 1-3% trailing revenue dip - follow-through failure tied to revenue
  • High-priority findings open >14d that reference checkout / payment / inventory problems the commerce sibling confirms are still active
  • Findings concentrated on one audit module (e.g. payment, inventory) the team cannot resolve - route to the owning connector for a structural fix

​
Severity thresholds

SignalWarnCritical
cri_vortexiq_findings_abandoned515
cri_vortexiq_finding_resolution_rate7550
cri_vortexiq_findings_open2040
cri_oldest_open_ticket_age3090
cri_unassigned_tickets520
cri_blocked_tickets38

​
Data sources

  • GET https://api.crisp.chat/v1/website/{website_id}/operators/list - Auth probe + operator inventory (team capacity)
  • GET https://api.crisp.chat/v1/website/{website_id} - Website sanity (name/domain) + tier check
  • GET https://api.crisp.chat/v1/website/{website_id}/conversations/{page_number} - List conversations - state counts, segments (priority + vortexiq:finding), assignee, ages
  • GET https://api.crisp.chat/v1/website/{website_id}/conversation/{session_id}/meta - Per-conversation meta - assigned operator, segments, timestamps
  • GET https://api.crisp.chat/v1/website/{website_id}/conversation/{session_id}/messages - Message timeline - last_activity_at, first-response/resolution timing, human-vs-bot detection