● Pre-launch (this week)
Tracking sprint
Provision Meta Pixel + System User token, Everflow offer + tracking domain, Cloudflare Worker secrets, Zaraz/GTM auto-inject. Then DNS cutover of degreesources.com. ~2-3 operator hours. See STATUS.md.
◑ First paid campaign
Daily 3-block rhythm
Morning Block (Dashboard Scan + Performance Review + Lead Quality + Social/Content + Alerts). Campaign Block (15-min brief → pipeline → review → export). Afternoon Block (platform-specific optimization rotating Mon-Fri).
◐ Second paid campaign
Weekly + monthly cadence
Weekly deep dives (M/W/F competitive intel + intel digest + A/B review). Monthly reviews (P&L + government data refresh + EDU event calendar planning).
○ Team-of-3 + 2+ verticals
Scaling roles + handoff
Solo → team-of-3 scaling roles + handoff guide for hiring + multi-vertical coordination. Future-quarter activation as expansion lands.
📖 The runbook below is the target-state operator manual. None of it is wrong; none of it is currently running daily because campaigns aren't live yet. Preserved for the moment first-dollar attribution lands — that's the day this becomes the operator's daily reference.
1
Daily Operating Rhythm -- 3 Blocks
Every day starts the same way regardless of verticals or campaigns. Morning = read, Midday = build, Afternoon = optimize.
5 min
Dashboard Scan
Active campaigns status → Yesterday's spend vs budget → CPL by platform → Creative fatigue alerts → Agent costs → System alerts
10 min
Performance Review
Per-angle metrics. Winners (CPL < target) → increase budget. Losers (CPL > 2x, 3+ days) → pause. Fatiguing (CTR declining) → queue creative refresh.
5 min
Lead Quality Check
Today's volume vs yesterday (drops = tracking issue). Quality tier distribution. State distribution. Buyer acceptance rate (Phase 5).
10 min
Social & Content Check
Posts pending approval? Engagement on yesterday's posts? Tool pages getting organic spikes? Respond to DMs/comments.
5 min
Alerts & Housekeeping
Competitive intel alerts. Regulatory changes. Coolify service health. Cron failures overnight.
15 min
New Campaign: Fill Brief
Partner, offer, audience, platforms, budget, compliance. Most fields have vertical defaults.
3 min
Stage 1-2: Validate + Strategy
Agent generates angles. Review: Vault winners (blue) = proven. Market (yellow) = data-driven. Original (red) = hypotheses. Approve 3-5 angles.
15 min
Stage 3 + QA: Copy Matrix
Full matrix generates in 5-10 min. Auto-QA scores. PASS (green) → approve. FLAGGED (yellow) → read carefully. FAIL (red) → auto-blocked.
10 min
Stages 4-7: Creative → Structure → Lander → Export
Review creative gallery. Generate assets or upload UGC. Build structure. Optional landing page. Export package. Upload to ad platforms.
30 min
Platform-Specific Optimization (Rotating Daily)
MON
Meta
Search terms, overlap, frequency
TUE
Google
Negatives, QS changes
WED
Content
Analytics, A/B, tools
THU
TikTok/Native
Creative, publishers
FRI
Cross-Platform
Blended CPL, attribution
2
Weekly Operating Rhythm -- M/W/F Deep Dives
Three structured deep-dive sessions per week. Each takes ~1 hour.
Monday: Strategy
Review last week's performance summary
Which verticals profitable? Which burning?
Review competitive intelligence
New competitor ads / angles to test?
Plan this week's campaigns
Budget reallocation decisions
Document in Plane
Wednesday: Content
Articles in draft/review status
Tools needing data refreshes?
Social calendar: schedule rest of week
Queue tool-highlight posts
Email: open rates, click rates (P3+)
A/B test subject lines
Friday: Retro
Campaign retros (7+ day campaigns)
Winners Vault auto-updated
Agent SDK cost audit
Creative gen costs trending
Coolify health: memory, disk
Metabase query performance
Cron failure review
Tue/Thu: Daily Ops
Morning read (30 min)
Campaign work if active gen
Platform optimization (rotate)
Social posting + engagement
Weekend: Off
Campaigns run on autopilot
Automated alerts only
No budget changes
Learning phases continue
3
Campaign Lifecycle -- Test → Scale → Maintain
Every campaign follows this progression. The 3-day rule: never kill before 3 full days of data.
TEST
$50-100/day
3-7 days
• Upload in PAUSED state
• Verify tracking end-to-end
• Highest Volume bid (Meta)
• Max Clicks (Google)
• Day 1-2: DO NOT TOUCH
• Day 3: first read
• Day 5-7: decision point
→
SCALE
$200-500/day
2-4 weeks
• +20-30%/day budget ramp
• Duplicate winners w/ variations
• Add new platforms
• Creative refresh for fatigue
• Bid transition: volume → CPA
• A/B test landing pages
• Add retargeting windows
→
MAINTAIN
Steady state
Ongoing
• Every 2 wk: check fatigue
• Generate new variants
• Test 1-2 new angles
• Update negatives (Search)
• Monthly: full retro
• Budget reallocation by ROAS
↻
REFRESH
As needed
When fatigued
• Same angle, new creative
• New hooks from Winners Vault
• Seasonal angle rotation
• Re-enter test phase
• Vault patterns → "always test"
• 3x losers → retire angle
Day 5-7 Decision Tree
Day 0
Generate → Upload PAUSED → Verify tracking → Publish
Day 1-2
LEARNING PHASE -- do not touch. Monitor for tracking errors and disapprovals only.
Day 3
First read: per-angle CPL and CTR. Pause clearly broken (0 conversions, 10x CPL). Note winners.
Day 5-7
WINNERS: CPL < target, consistent → move to SCALE phase
BORDERLINE: CPL 1-1.5x target → 3 more days with creative refresh
LOSERS: CPL > 2x target, no trend → kill. Free budget for winners.
4
Content-to-Social Pipeline
When content publishes, n8n auto-generates social variants. You approve in /brand/queue. Video platforms flagged for manual creation.
Automated (n8n → queue → approve)
Manual (Plane task created)
5
Monthly Operating Rhythm
Three monthly checkpoints: data refresh + planning, optimization deep dive, financial review.
First Week: Data & Planning
Government data refresh (verify cron results)
College Scorecard, CMS, state grants
Update tool parameters if data changed
Monthly performance report (all verticals)
Vertical expansion review: stable? expand?
Content calendar for next month
Mid-Month: Optimization
Winners Vault review -- patterns emerging?
Promote strongest patterns to "always test"
Retire 3x losers permanently
A/B test audit (GrowthBook significance)
Implement winners, start new experiments
Partnership pipeline update + outreach
End of Month: Financial
P&L: revenue vs costs by vertical
Infrastructure costs ($56/mo DO + APIs)
Agent SDK costs trending
Net margin by vertical
Budget decisions: increase/pause verticals
Update STATUS.md + Plane tasks
6
Team Scaling -- Solo → 3 People
Start solo. Hire Person #2 when running 3+ campaigns across 2+ platforms. Person #3 when content volume exceeds solo review capacity.
You -- Solo Operator
Month 1+ • 3-4 hours/day
Campaign generation (agent does work, you decide)
Performance monitoring (30 min/day)
Content publishing + social (30 min/day)
Lead quality monitoring (15 min/day)
Competitive intel (15 min/week)
Brand management (30 min/day)
Infrastructure (as needed)
Person #2: Media Buyer
Month 3+ • When 3+ campaigns x 2+ platforms
Daily performance monitoring
Copy/creative approval (delegated, trained)
Platform-level optimization (bids, audiences, negatives)
A/B test management (GrowthBook)
Social posting (delegated)
Does NOT: generate campaigns, make compliance decisions, approve new angles, touch offer distribution.
Person #3: Content & Research
Month 6+ • When content > solo review capacity or 3rd+ vertical
Article writing/editing (/content-writer)
Tool content updates (gov data refresh)
Vertical research (Firecrawl sessions)
SEO monitoring (rankings, content gaps)
Partnership outreach
Does NOT: manage ad spend, touch pipeline generation, make budget decisions.
7
EDU Event Calendar -- Seasonal Content Triggers
Tie social posts and campaign pushes to real-world events. These are the moments that drive traffic and urgency.
Sep-Oct
FAFSA opens Oct 1
"Use our EFC Calculator before you file"
Jan-Mar
State FAFSA deadlines
Geo-targeted deadline alerts
Cal Grant: March 2
Mar-May
Award letter season
"Use our Aid Letter Decoder"
Decision deadlines
May-Aug
Enrollment decisions
"ROI Calculator shows real numbers"
Year-round: rotate all 10 tools
5
Weekly Operating Rhythm
Three focused blocks per week. Each 1 hour. Monday strategy, Wednesday content, Friday retro.
Monday: Strategy & Planning (1 hr)
1. Review last week’s campaign performance summary (/analytics weekly roll-up)
2. Which verticals profitable? Which burning cash? Hitting daily spend targets?
3. Review competitive intelligence (/dashboard Competitive Radar card)
4. New competitor ads: any angles we’re not running? LP changes to test?
5. Plan this week’s actions: new campaigns, creative refreshes, budget reallocation
6. Document decisions in Plane
Wednesday: Content & Brand (1 hr)
1. Content pipeline check: articles in draft/review, which ready to publish?
2. Any tools need data refreshes? (government data changes)
3. Social content planning: schedule posts for rest of week
4. Queue tool-highlight posts (drive free traffic to high-converting pages)
5. Review engagement metrics from this week’s posts
6. Email performance (Phase 3+): open rates, click rates, A/B subject lines
Friday: Performance & Retro (1 hr)
1. Run /campaign-retro on any campaign with 7+ days of new data
2. Agent analyzes: winners, losers, fatigued angles → Winners Vault updated
3. Cost audit: Agent SDK spend, creative generation costs this week
4. Are we getting cheaper as the vault fills? (should be)
5. Infrastructure health: Coolify memory, disk, Metabase query performance
6. Scheduled task check: any cron failures this week?
6
Campaign Lifecycle — Test → Scale → Maintain
Every campaign follows the same 3-phase pattern. The numbers change, the process doesn’t.
Test Phase
$50-100/day • 3-7 days
Day 0: Generate + upload (PAUSED)
Day 1-2: Learning phase — DO NOT TOUCH
Day 3: First read — pause clear losers
Day 5-7: Decision — scale winners, kill losers
Rule: Never kill before 3 full days
Scale Phase
$200-500/day • 2-4 weeks
W1: Ramp 20-30%/day, duplicate winners
W2-3: Creative refresh, audience expansion
Bid transition: Volume → Cost Per Result
W4: Steady state — CPL stable at scale?
Rule: Never increase >20%/day
Maintain Phase
Ongoing • Every 2 weeks
Check creative fatigue (CTR decline + frequency rise)
Generate new variants for winning angles
Test 1-2 new angles from latest strategy
Monthly: full retro, vault update, budget realloc
Rule: Change execution, not the angle
7
Incident Response Playbooks
Four scenarios. Diagnose → Fix. Don’t panic — follow the playbook.
Tracking Failure HIGH IMPACT
Symptoms: Zero conversions, leads not in Everflow
Check: Zaraz firing? → Everflow receiving? → CAPI Worker errors? → LP pixel present?
Fix: Usually stale token or misconfigured pixel ID. Verify with test conversion.
Campaign Disapproval MEDIUM
Symptoms: Ad disapproved, campaign paused
Check: Read disapproval reason → check compliance config → check LP for policy violations
Fix: Edit violation, resubmit. Do NOT appeal unless certain it’s false positive.
Agent Execution Failure LOW
Symptoms: Pipeline stage fails
Check: Which stage? → agent_sessions.error → API error or data error?
Fix: Re-run failed stage. If fails again, check skill files for changes.
Cost Overrun LOW
Symptoms: Agent cost card high
Check: /analytics/costs → which session spent most? Runaway loop? Large context?
Fix: Set per-session cost cap ($10 hard limit) in agent executor.
8
Scaling: Solo → Team of 3
When to hire, what they own, what they don’t touch.
Solo (Month 1+)
You do everything.
3-4 hrs/day active time.
Agent handles generation.
You handle decisions, approvals, optimization.
+Media Buyer (Month 3+)
Hire when: 3+ campaigns, 2+ platforms, can’t monitor closely enough.
Owns: daily monitoring, platform optimization, copy approval (trained), social posting.
NOT: campaign generation, compliance, new angles, offer distribution.
+Content Operator (Month 6+)
Hire when: content volume outpaces review, or 3rd+ vertical.
Owns: article writing/editing, tool data updates, vertical research, partnership outreach.
NOT: ad platforms, infrastructure, financial data.
Training Protocol (Same for Both Hires)
Week 1: Observe your routine. Learn the dashboard. Read compliance rules.
Week 2: Supervised action. They propose, you approve/deny. Daily 15-min sync.
Week 3: Independent with guardrails. They own their domain, you retain strategic decisions.