11 search terms above $200 with 0 conversions (total: $4,679)

• › Add top wasters as negative keywords in Nordanyan Law shared negative list
• › Review Nordanyan Law match types — switch broad match to phrase where waste is highest

Zero calls during business hours: 2 PM

• › Review Nordanyan Law call_volume trend over the past 7 days for pattern
• › Check if any recent changes (bids, budgets, targeting) correlate with this alert

PACING ALERT: Nordanyan Law WTD conversions -33.3% vs same point last week (2 vs 3)

• › Compare Nordanyan Law WTD conversion volume to spend — is CPA rising or traffic dropping?
• › Check Nordanyan Law search term quality and keyword performance for the week

SPEND SPIKE: Nordanyan Law spend up 103% WoW ($4,612 -> $9,378)

• › Review Nordanyan Law campaign-level spend breakdown — identify which campaign spiked
• › Check if Nordanyan Law bid strategy or budget was recently changed

Spend pacing :arrow_down: 56% under (projected $17,581 vs $40,000 budget)

• › Review Nordanyan Law daily budget caps and campaign-level spend limits
• › Check if any Nordanyan Law campaigns were paused or have limited serving

No conversion uploads detected in window — upload pipeline may be stalled

• › Review Nordanyan Law conv_monitor trend over the past 7 days for pattern
• › Check if any recent changes (bids, budgets, targeting) correlate with this alert

AI Ops Weekly Digest — Week of July 27, 2026 ============================================================ COST STATUS: ON TRACK MTD: $662.58 | Projected: $760.74 | Budget: $1,500.00 (51%) 7-day avg: $27.94/day MODEL EFFICIENCY: Opus ratio: 100% | Wasteful sessions: 5 | Est. waste: $151.81 COST ALLOCATION: RGDM (internal): $18,437.94 (28 sessions) SESSION QUALITY: HEALTHY Avg score: 98/100 | Cache hit: 93.5% | Flagged: 0/28 OVERHEAD TREND: STABLE Latest avg input: 14,495,140 tokens/session | WoW: +39.0% | Consecutive growth: 2 weeks RECOMMENDATIONS: 1. [HIGH] Route trivial sessions (<5 min) to Sonnet/Haiku instead of Opus (est. $607.24/mo savings) ALERTS: [MEDIUM] (efficiency) Opus ratio 100% exceeds 25% threshold [MEDIUM] (efficiency) Estimated $151.81 waste from 5 trivial Opus sessions [LOW] (allocator) RGDM (internal) accounts for 100% of AI spend — consider cost optimization OUTCOME TRACKING: [-] Opus Ratio: 100% (was 93%, +7.9% — regressed) [+] Est. Waste: $151.81 (was $152.66, -0.6% — improved) [-] Budget Pacing: 51% (was 48%, +5.0% — regressed) [+] Avg Quality: 98/100 (was 94/100, +4.9% — improved) [+] Cache Hit: 93.5% (was 85.2%, +9.7% — improved) [-] Overhead: 14,495,140 tok (was 10,429,828 tok, +39.0% — regressed) 4-week trends: Budget Pacing: 24% → 20% → 22% → 48% (worsening) Opus Ratio: 100% → 100% → 100% → 93% (improving)

• › Route trivial sessions (<5 min) to Sonnet/Haiku instead of Opus