Skip to main content
Nerve Centre KPIs · Audit Profile · Sentiment Settings Help Scout conversation state means little to a merchant unless it’s joined to the audit findings VortexIQ filed and to the revenue those fixes protect. This audit answers: (1) is the OAuth token still good and the mailboxes readable, (2) is the backlog under control (oldest open, blocked, by priority/status), (3) are people overloaded or work left unassigned, and (4) - the stickiness test - are the findings we filed into Help Scout actually getting resolved, or quietly dying in the queue (abandoned >14d, resolution rate <50%)?

​
What this audit checks

​
Authentication & access

  • OAuth token valid (auth on /v2/users/me) and not expired/revoked
  • client_credentials grant still issues a token (App ID/Secret current)
  • Connector app has Conversations + Reports read scope
  • All expected mailboxes readable (multi-mailbox aggregation complete)

​
Backlog health

  • Open conversation count vs 30-day baseline (sudden growth = drowning signal)
  • Blocked / waiting-tagged pending conversation count + change WoW
  • Oldest open conversation age (>90 days = backlog rot)
  • Backlog-by-priority skew (Highest+High priority tags >50% of open = firefighting)
  • Backlog-by-status distribution (active/pending/closed/spam mix)

​
Throughput & capacity

  • Conversations resolved (7d) dropped >25% vs prior week
  • Avg cycle time degradation >25% week-over-week
  • Assignees with >10 open conversations (overload signal)
  • Unassigned conversations > 5 (warn) / > 20 (critical) - work nobody owns
  • Sprint scope-creep (>15% added after the triage window started)

​
VortexIQ findings lifecycle (the stickiness test)

  • Abandoned findings - vortex_iq-tagged conversations, no status change >14d
  • Finding resolution rate over rolling 90d (target >75%, alert <50%)
  • Avg time-to-fix for resolved findings (>14d sustained)
  • Findings closed without a deploy/evidence note (silent dismissals)
  • Regression rate - findings reopened after closed

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

  • Critical findings from sibling connectors (shopify/bigcommerce/adobe/cybersource) older than 7 days with NO Help Scout conversation - coverage gap
  • Abandoned findings count rising while a commerce sibling’s refund_rate climbs - measured-but-not-fixed revenue leak
  • High-LTV customers (top decile per commerce sibling) waiting on unresolved vortex_iq conversations
  • Help Scout conversations referencing decommissioned / archived integrations - stale work

​
Severity thresholds

SignalWarnCritical
vortexiq_findings_abandoned515
vortexiq_finding_resolution_rate7550
blocked_tickets38
unassigned_tickets520
oldest_open_ticket_age_days6090
cycle_time_change_pct2550
scope_creep_pct1530

​
Data sources

  • GET https://api.helpscout.net/v2/users/me - Token + account/region sanity
  • GET https://api.helpscout.net/v2/mailboxes - Mailbox inventory for multi-mailbox aggregation
  • GET https://api.helpscout.net/v2/users - Agent inventory + capacity calc (assignee load)
  • GET https://api.helpscout.net/v2/tags - Tag taxonomy (priority/blocked/topic clustering)
  • GET https://api.helpscout.net/v2/conversations - Backlog reads: status/tag counts, oldest-open, by-assignee, abandoned-findings, blocked
  • GET https://api.helpscout.net/v2/reports/conversations - Throughput / volume / closed aggregates (resolved-7d, trend, velocity)
  • GET https://api.helpscout.net/v2/reports/productivity - Resolution time, replies, handled (cycle/lead time)
  • GET https://api.helpscout.net/v2/reports/company - Resolved counts over windows (finding resolution-rate numerator/denominator, time-to-fix)