HiON
Stop-lossGoogle CPL crossed $500 for the 4th consecutive week· Google CPL this week: $651 (4 leads at $2,603 spend). Recommendation: pause current creative, test a new angle.
Investigate →
Stop-lossMeta CPL crossed $400 for the 13th consecutive week· Meta CPL this week: $958 (4 leads at $3,832 spend). Recommendation: pause current creative, test a new angle.
Investigate →
HiON · FranDev · System health & ops

Is the plumbing working?

Daily glance — most days the team doesn’t think about it. When something feels off, this is the first place to look.

Synthesized · Claude narrative
Confidence · High

Plumbing is up. Form-to-Monday SLA holding at p90 3.4s (target 5s). Monday-to-Bland AI is the soft spot — p99 spikes to 41s on cold starts; webhook retry handled it correctly in the last 24h. Data quality clean except 1 zombie candidate (>60d in Education) that needs a manual decision. Last failed automation was 6h ago — Zapier transient 502 from Monday API, auto-retried successfully. Bland AI script v3.2 (deployed May 8) is the current version.

01Automation pipeline
Form → Monday → Bland AI → Slack. Each node shows last-successful event and status.
01
Landing page form
3m ago
02
Monday Board 1
3m ago
03
Bland AI agent
2m ago · p99 outlier
04
Monday update
2m ago
05
Slack notification
2m ago

Node turns amber if last-successful-event timestamp is older than the SLA window for that step; red if the step has hard-failed in the last hour. Bland AI is amber due to a p99 cold-start outlier — auto-retry handled it.

02SLA compliance
Five SLAs from the Monday plan. Each bar shows p50 / p90 / p99 vs target.
  • Form submit → Monday item
    p50 1s · p90 3s · p99 5s · target 5s
  • Monday item → Bland AI call init
    p50 8s · p90 22s · p99 41s · target 30s
  • AI call connects (form → connect)
    p50 28s · p90 56s · p99 4.7m · target 1.0m
  • Bland AI ends → Monday update
    p50 2s · p90 4s · p99 6s · target 5s
  • SQL flagged → Slack ping
    p50 2s · p90 9s · p99 27s · target 30s
03Data quality inventory
Records missing required fields or stuck past sanity thresholds. Click to land on a filtered Monday view (Phase 3 polish).
04AI agent performance
Bland AI. Calls per hour by day reveals whether off-hours calling is hurting answer rate.
Answer rate
85%
calls connected / attempted
Completion rate
72%
all qualifying Qs answered
Avg call duration
184s
time on call when connected
Calls this period
118
v3.2 · deployed May 8
Calls by hour-of-day · weekdays
8
9
10
11
12
13
14
15
16
17
18
19
Mon
2
4
4
7
7
7
7
4
4
4
2
2
Tue
1
3
3
6
6
6
6
3
3
3
1
1
Wed
1
3
3
6
6
6
6
3
3
3
1
1
Thu
1
3
3
6
6
6
6
3
3
3
1
1
Fri
1
3
3
6
6
6
6
3
3
1
0
0
05Audit log
Last 15 automated actions. Filter by actor (Phase 3 polish: full 100 + filters).
  • 11dZapierWebhook: form submit → Monday create_item (lead-1023)
  • 11dBland AIInbound call initiated for lead-1023 (3.1s after submit)
  • 11dBland AICall completed · score 8 · duration 168s
  • 11dMondaylead-1023 Stage → Qualified SQL, Slack notification queued
  • 11dSlackSent: '#sales-handoffs · Sarah Chen (lead-1023, score 8)'
  • 11dCronDaily snapshot rolled up to Board 3 (2026-W21)
  • 11dBland AICall failed · no answer · enrolled in retry queue
  • 11dZapierCalendly event → Monday update (cand-0012 DD booked)
  • 11dMondayStale candidate alert: cand-0001 > 14d in Education
  • 11dClaudeWeekly synthesis generated for 2026-W21
  • 11dBland AICall connect latency 312s (p99 outlier, cold start)
  • 11dSlackStop-loss notification: Meta CPL crossed $400 (9th day)
  • 11dCronWeekly metrics rollup completed
  • 11dMondayLead promotion automation: 3 SQLs → Active Candidates
  • 11dZapierBland AI webhook retried (transient 502 from Monday API)