👤 Human Review Gates

BaC Principle

“Humans should direct the process, not execute it. Every gate is a decision point — not a work session.” — Business as Code Manifesto

This file defines exactly where humans enter the SEO & AEO process and what decisions they make. Agents handle execution. Humans handle judgment calls — strategy alignment, brand consistency, factual accuracy.

See Map of Content for full vault navigation.

---

🗂️ Gate Overview

Gate	Stage	Input	Time Required	Decisions
Gate #1 — Strategy Review	After S2	Keyword clusters + content calendar	~20 min/month	APPROVE / REPRIORITIZE / REJECT_CLUSTER
Gate #2 — Content Review	After S4+S5+S6	10% sample of optimized pages	~30 min/week	APPROVE / EDIT / PAUSE_BATCH / FLAG

BaC Result

Total human time: ~60 min/week including both gates and the improvement loop review in 10-Metrics-and-Self-Improvement.

---

🚦 Gate #1 — Strategy Review

Trigger: 04-Keyword-and-Topic-Research-Agent delivers keyword_topic_map Stage: S3 Frequency: Monthly (or when a new topic cluster batch is ready) Time budget: ~20 minutes

Input Package

The agent delivers a pre-compiled strategy brief containing:

• Ranked list of keyword clusters (by score)
• Recommended content calendar (next 4 weeks)
• AEO query map (what AI models currently answer for these topics)
• Content gap summary (what’s missing vs. what exists)
• Any clusters flagged as “borderline” by the agent

Decision Framework

What you’re looking for	Decision
Clusters align with business goals and current company priorities	APPROVE — proceed to content optimization
Good clusters, but wrong order — some should ship sooner	REPRIORITIZE — reorder and approve
New topic the agent missed — relevant to our audience	ADD_TOPIC — add to cluster list
Cluster doesn’t match what we sell or who we serve	REJECT_CLUSTER — remove and note reason
Scoring model seems off — too many irrelevant suggestions	FLAG_MODEL — trigger Target Definition review

Approval Heuristics (15-min review guide)

1. Scan top 10 clusters by score (2 min)
   → Do these feel right? Are there obvious misfits?

2. Check content calendar sequence (3 min)
   → Is the order logical? Does the first piece establish context for later ones?

3. Review AEO query map for top 5 clusters (5 min)
   → Are these questions our audience actually asks?
   → Is our site positioned to answer them credibly?

4. Spot-check 2–3 "borderline" clusters flagged by agent (5 min)
   → Are these worth investing content effort in?

5. Make decisions and record in feedback log (5 min)

Approval Threshold

gate_1_threshold:
  approve: "≥70% of clusters approved without major changes"
  revise:  "30–70% approved → reorder and return to agent for re-run"
  reject:  "<30% approved → flag [[02-Target-Definition]] for revision"

Feedback Format

{
  "gate": "strategy_review",
  "reviewed_by": "CMO",
  "reviewed_at": "2026-04-02",
  "decisions": [
    {
      "cluster_id": "C001",
      "decision": "APPROVE",
      "priority": 1,
      "notes": ""
    },
    {
      "cluster_id": "C004",
      "decision": "REJECT_CLUSTER",
      "reason": "Too generic — not aligned with BaC positioning",
      "notes": "Consider replacing with 'AI-executable SOP' angle"
    }
  ],
  "calendar_adjustment": "Move C007 to week 1 — timely topic",
  "new_topics_added": ["AI agents for operations teams"],
  "overall_decision": "APPROVE_WITH_CHANGES"
}

---

🚦 Gate #2 — Content Review

Trigger: 05-Content-Optimization-Engine + 06-AEO-Structuring-Agent + 07-Technical-SEO-Agent deliver approved_publish_batch Stage: S7 Frequency: Weekly (each content batch) Time budget: ~30 minutes

Input Package

The agent delivers:

• 10% random sample of the content batch (or all pages with AEO score < 75)
• Per-page summary: word count, SurferSEO score, AEO score, schema status, changes made
• Flagged pages: unsourced claims, AEO score misses, schema failures
• Technical fix specs from 07-Technical-SEO-Agent (for awareness, not action)

Decision Framework

What you’re seeing	Decision
Content is accurate, brand-consistent, AEO structure intact	APPROVE — publish batch
Minor issues in sample — small edits needed	EDIT_AND_RESAMPLE — edit flagged pages, re-sample from batch
Significant quality issues in sample	PAUSE_BATCH — do not publish; return to content engine
One specific piece has a factual error or brand risk	FLAG_PIECE — remove that page from batch; publish rest

Quality Checklist (per sample page)

content_review_checklist:
  accuracy:
    - "All factual claims are correct to your knowledge"
    - "No outdated statistics or superseded information"
    - "No hallucinated citations or non-existent source links"
 
  brand:
    - "Tone is authoritative but accessible — not academic, not casual"
    - "BaC terminology used consistently"
    - "No competitor names used in unflattering ways"
 
  aeo_structure:
    - "First paragraph of each major section answers its heading's question directly"
    - "FAQ section is present and questions are phrased naturally"
    - "Key terms are defined on first use"
 
  seo:
    - "Title and meta description present and relevant"
    - "H1 is present and matches topic"
    - "Internal links are contextual and anchor text is descriptive"
 
  technical:
    - "Schema markup is present (visible in page source or confirmed by agent)"
    - "No broken links in the content"

Approval Threshold

gate_2_threshold:
  approve:         "Sample passes checklist with 0 critical issues"
  edit_resample:   "1–2 minor issues in sample → edit + re-sample"
  pause_batch:     "≥3 critical issues OR any hallucinated facts"
  flag_piece:      "One page has brand/accuracy issue → remove, publish rest"

Feedback Format

{
  "gate": "content_review",
  "reviewed_by": "CMO",
  "reviewed_at": "2026-04-09",
  "batch_id": "batch_2026-04-07",
  "pages_in_batch": 8,
  "pages_sampled": 1,
  "decisions": [
    {
      "page_id": "page_C001",
      "url": "/what-is-business-as-code",
      "decision": "FLAG_PIECE",
      "reason": "Stat in paragraph 3 is from 2022 — needs updating",
      "action": "Update with 2025 data from Gartner report"
    }
  ],
  "batch_decision": "APPROVE_WITH_EXCEPTIONS",
  "flagged_pages": ["page_C001"],
  "publish_remaining": true,
  "feedback_to_agents": {
    "content_optimization_engine": "Ensure statistics are <2 years old",
    "aeo_structuring_agent": "FAQ on page C003 has duplicate questions — tighten deduplication"
  }
}

---

⏱️ Weekly Time Budget

Activity	Time	Frequency
Gate #1 — Strategy Review	20 min	Monthly (1× per month)
Gate #2 — Content Review	30 min	Weekly
Metrics & Improvement Review	30 min	Weekly
Total per week (avg)	~65 min

When to escalate

If Gate #2 is consistently surfacing the same issue (e.g., outdated stats, wrong tone), escalate to the agent prompt — not the gate. The gate should be catching exceptions, not patterns. Patterns belong in 10-Metrics-and-Self-Improvement.

---

📎 Related Files

• 04-Keyword-and-Topic-Research-Agent — Delivers input for Gate #1
• 05-Content-Optimization-Engine — Delivers content for Gate #2
• 06-AEO-Structuring-Agent — AEO scores inform Gate #2 sample selection
• 07-Technical-SEO-Agent — Technical specs reviewed at Gate #2 (awareness)
• 10-Metrics-and-Self-Improvement — Gate feedback feeds improvement proposals
• 01-Process-Manifest — S3, S7 stage definitions