AEO Content

Produce articles citation-worthy by ChatGPT, Claude, Perplexity, and Google AI Overviews using the CITABLE framework + retrieval-citation-trust pipeline. Two output modes: CITABLE block (800-1,500 words) for definition/FAQ/short BOFU pages, and authority long-form (2,000-4,000 words) for competitor comparisons, listicles, how-to guides, and industry guides. Review gate is Level 3 because output publishes externally under client domain.

Doctrine inherited (Step 7 — 0626 rollout)

Output complies with:

• output-tenets.md — the seven tenets
• output-simplicity.md — length caps, three-layer source placement, robot-tells ban
• doc-output-structure.md — GDoc/Notion structural defaults (when published to either before site)
• Step 6 calibration: see [[feedback_execution_doctrine_refinements_step6]]

Refinements applied to this skill:

Code	Refinement	How it lands in aeo-content
R1	Source placement (three layers)	Published articles are end-customer-facing. No "Sources" block on the rendered page — citations live in working draft for QA only, then convert to inline links inside the prose at publish-time. The page itself never carries a sources block; that's an AI-citation tell.
R3	Product-update tone	Any product / feature mention frames as "[Product] does X" not "we are thrilled to announce [Product]." Even hero comparison pages stay even-keeled.
R6	CTA hierarchy	TOFU/MOFU articles (definition, FAQ, how-to, industry guide) → blog-CTA + sign-up secondary. BOFU articles (competitor comparison, listicle ending in vendor pick) → sign-up or trial primary.
R7	FAQ titles + no sources block	FAQ articles use canonical FAQ titles ("What is [X]?", "How does [X] work?", "How to [task]"). Never marketing-flavored variants ("X in 60 seconds," "The ultimate guide to X"). No sources block, two-types pattern (educational longer / how-to ≤4 bullets).
R8	Entity-name headings	Section headings repeat the entity name — "What [Product] does," "How [Product] is different" — not pronoun headings ("What it does"). Customers scan for the entity.
R9	Action-oriented section names	"How to set up [X]" beats "Setup." "Why [X] matters" beats "Importance." "How [X] compares to [Y]" beats "Comparison."

When to run

• After aeo-strategy produces an article queue (Mode B — title, keywords, cluster, funnel stage, mode supplied).
• Standalone for a single article (Mode A — user provides topic + 3+ target queries).
• After landing-page-copy finishes (auto-suggest: AEO content can extend a landing page into a citable hub).

Inputs

Required: topic/concept; 3+ target queries (primary + adjacent); company context (capabilities, differentiators). Mode B from aeo-strategy queue: article title, target keywords, cluster, funnel stage, content mode — skip Phase 1. Optional (improve quality): ICP research, competitor content, existing proof points, current page (for refresh), competitor research deep dives. Client context: auto-loaded from client CLAUDE.md — apply Voice & Messaging rules, vocabulary, anchors automatically. Validation: topic specific enough; ≥3 queries; company context available; mode determined. Full input matrix + validation checklist: the premium reference.

Steps

1. Validate inputs. Confirm topic, queries, company context. If missing, ask user or offer Exa research per .claude/rules/exa-protocol.md.
2. Phase 1.1 — Map queries to content type. Use query pattern table in the premium reference ("[X] vs [Y]" → comparison; "What is [X]" → definition; "How to [X]" → problem-solution; "FAQ" or multi-question → FAQ).
3. Phase 1.2 — Identify 5-7 adjacent intents. Patterns: What/How/Why/vs/best/examples. These become H2 sections.
4. Phase 1.3 — Load template. From the premium reference (comparison-page.md, definition-page.md, problem-solution-guide.md, faq-page.md, authority-long-form.md, bofu-comparison.md, mofu-listicle.md, mofu-how-to.md, content-refresh.md). Confirm structure with user.
5. Phase 2 — Apply CITABLE (7 letters). C: BLUF (40-80w) + key facts box (3-5 sourced); I: answer primary in first 100w + H2s for adjacent; T: 3+ third-party sources, G2/Capterra if relevant; A: 1-2 quotable facts/section, inline [Source, Year]; B: 200-400w sections, TL;DR, comparison tables (<10 rows), jump links if >1,500w; L: visible "Last updated: YYYY-MM-DD" + refresh schedule; E: relationship statements + schema plan. Per-letter deliverables: the premium reference.
6. Phase 3.1 — Add structural elements. TL;DR box, ≥1 table, FAQ (5-10 questions), hub-and-spoke internal links, schema markup plan.
7. Phase 3.2 — Apply citation-boost stack. FAQ +40%, comparison table +30-40%, numbered list +25%, TL;DR +20%, key facts box +20%. Verify each present.
8. Phase 3.3 — Generate schema. JSON-LD for FAQPage/HowTo/Product. Validate schema matches visible content exactly (anti-hallucination rule 3).
9. Self-evaluation gate. All facts sourced? Citations real (URLs verified)? Schema matches visible? No invented stats/quotes/G2 ratings? If fail: use [PLACEHOLDER] or [NEED TO VERIFY].
10. Format output. CITABLE block uses template in the premium reference; authority long-form uses templates directly. Inter font, sentence-case headers, [Source, Year] citations, max 10 row tables.
11. Run pre-delivery checklist. CITABLE audit + format quality + completeness — see the premium reference.
12. Present at Review Gate 3 (deep review, external publication). Actions: [Approve] [Add sources] [Expand].
13. Post-output: offer iteration prompts (refinement, expansion, quality) — see the premium reference.
14. Capture learnings. Log feedback signals; on user approval, save to examples/[date]-[content-type].md per the reference-capture format.
15. Suggest chain. → lifecycle-marketing (content feeds nurture), landing-page-copy (AEO enhances LPs), competitor-research (deeper comparison content).

Full phase-by-phase walkthrough with checkpoints + flowchart: the premium reference. Mode + content-type mapping (BOFU/MOFU/TOFU): the premium reference.

What good looks like

Evaluations

• BLUF ≤100 words and answers primary query directly?
• Key facts box has 3-5 data points each with [Source, Year]?
• H2s cover 5-7 adjacent intents (not just primary)?
• Every section has 1-2 quotable facts with inline citations?
• ≥1 comparison table AND ≥1 numbered list present?
• FAQ has 5+ self-contained answers (40-60w direct + 100-200w expanded)?
• "Last updated: YYYY-MM-DD" timestamp visible at top?
• Schema JSON-LD matches visible content exactly (no hidden info)?
• Hub-and-spoke internal links identified?
• Zero invented stats/quotes/G2 ratings (all sources verifiable or marked [PLACEHOLDER]/[NEED TO VERIFY])?
• Output header includes skill name, date, font (Inter), version, content type?
• Citations use inline [Source, Year] format consistently?
• TL;DR matches mode (bullet list for CITABLE block, blockquote paragraph for authority long-form)?
• Refresh schedule defined and content-type-appropriate?

Sourced patterns — comparison page architecture

For comparison / alternative / "vs." pages (alt page format, plural alternatives format, you-vs-competitor format, competitor-vs-competitor format):

• "Honesty Builds Trust" pattern. Acknowledge competitor strengths and your own limitations explicitly. Reads as referee, not biased advocate. Trust-as-conversion.
• "Who it's for" sections per option. On a plural alternatives page (e.g., "Best alternatives to X"), each listed alternative gets a "Who it's for" block — explicit ICP fit per option. Helps evaluators self-select; converts at higher rate than positioning-against-all.
• 4–7 genuine alternatives on plural pages. Fewer reads as biased; more reads as overwhelming. The 4–7 range optimizes for "this is helpful" perception.
• TL;DR-then-depth hierarchy. Lead each option with a 1-sentence summary; follow with paragraph-level depth. Scanners take the TL;DR; deep readers go further.
• Conversion benchmark per source: comparison / alternative pages convert at 5–15% vs. 0.5–2% for generic content. The format earns its place when the cluster exists.

---

Final ship gate

Run /premortem --output before ship. See /premortem skill for the 5 execution domains (will-it-resonate / will-it-convert / will-it-stay-on-brand / will-stakeholder-push-back / will-it-degrade-over-time) and output template.

Then run /voice-reviewer — the content ship gate: voice + brand quality (pm-loop.md).

Trivial-case escape: ## Premortem\nNo failure modes — trivial change satisfies the contract for genuinely trivial outputs.

---

Persuasion & stickiness pass

Output complies with persuasion-and-stickiness.md — Cialdini's 7 persuasion levers + Heath's SUCCESs. Deploy the 1-2 Cialdini levers that fit the reader's barrier (never all seven; every lever must be TRUE), run the SUCCESs diagnostic (Simple / Unexpected / Concrete / Credible / Emotional / Stories) over the near-final draft, then the rule's pre-ship gate.