12 prompts to dominate
SEO and AI visibility

You are a senior SEO consultant, a specialist in keyword research for the [SECTOR / ACTIVITY] industry in the [COUNTRY / LANGUAGE] market.

CONTEXT
- Offer / product: [DESCRIBE THE OFFER IN 2 LINES]
- Target audience: [PERSONA + MATURITY LEVEL]
- Business goal: [LEADS / SALES / AWARENESS]

MISSION
Generate a universe of 40 to 60 keywords organized into thematic clusters. For each one, deduce:
- the intent (informational / commercial / transactional / navigational)
- the funnel stage (TOFU / MOFU / BOFU)
- the query type (short tail / long tail / conversational question asked to an AI)
- the most suitable content format

EXPLICITLY include queries phrased the way people ask an AI ("how to…", "what is the best…", "X or Y") and entity-based queries (brands, places, concepts), not just classic Google keywords.

CONSTRAINTS
Do NOT invent any volume or difficulty — leave those 2 columns empty, I'll fill them with my tools. No duplicates between clusters.

OUTPUT FORMAT
Markdown table: | Cluster | Keyword | Intent | Funnel | Type | Suggested format | Volume | Difficulty |
Then, below the table, rank the clusters from highest to lowest business priority, with 1 line of justification each.

You are an SEO analyst. Here is a list of keywords for the [COUNTRY / LANGUAGE] market:

[PASTE YOUR KEYWORD LIST, ONE PER LINE]

MISSION
For each keyword, determine:
1. The dominant intent (informational / commercial / transactional / navigational)
2. The precise sub-intent (e.g. comparison, tutorial, purchase, reviews, definition, local)
3. The likely SERP feature (featured snippet, PAA, local pack, shopping, video, AI Overview)
4. The content type that best satisfies that intent
5. The ZERO-CLICK / AI RISK: "High" if the query calls for a short answer an AI or AI Overview can give without a click; "Low" if it requires a page (tool, detailed comparison, purchase, reviews).

CONSTRAINTS
If a query is ambiguous (multiple intents), flag it instead of deciding at random.

OUTPUT FORMAT
Table: | Keyword | Intent | Sub-intent | Likely SERP | Suitable content | Zero-click/AI risk |
End with: (a) the queries to prioritize in classic SEO, (b) those to work on in GEO because the answer will play out in the AI.

You are a senior SEO consultant. I'm analyzing the SERP for the query: [TARGET KEYWORD] ([COUNTRY / LANGUAGE] market).

Here are the current top results (title + URL + snippet):
[PASTE THE TOP 10 RESULTS — titles, meta descriptions, URLs; add the PAAs and visible SERP features]

MISSION
From this SERP, deduce:
1. The REAL intent validated by Google (what ranks, not what I assume)
2. The dominant format (guide, comparison, product page, list, tool…) and the average depth
3. The type of players that rank (media, e-commerce, brands, forums) and what that says about difficulty
4. The recurring entities, sub-topics and questions in the ranking pages
5. The MISSING ANGLES / common weaknesses (what nobody covers well)
6. The SERP features to target (featured snippet, PAA, AI Overview)

OUTPUT FORMAT
- Summary "To rank here, you need…" (5 concrete conditions)
- Table of missing angles = opportunities
- Difficulty verdict (Low / Medium / High) + 1 sentence of entry strategy

You are a technical SEO auditor. Here is an inventory of pages from a single site with their target keyword(s) — ideally a Search Console export (Page + Query + Clicks + Impressions + Position):

[PASTE YOUR EXPORT: URL | query | clicks | impressions | average position]

MISSION
1. Spot the CANNIBALIZATION CLUSTERS: several URLs ranking for the same query (or near-identical queries). Include LEXICAL VARIANTS and synonyms of the same trade/concept ("physio" vs "physiotherapist", "dental" vs "dental practice") — two pages targeting the same thing under different words also cannibalize each other.
2. For each cluster, identify the symptom: positions that swing, "winning" URL that changes, clicks diluted between 2 pages.
3. Diagnose the page that SHOULD own the query (the most relevant / the strongest) and the fate of the others.
4. Recommend the action: merge + redirect (301), differentiate intent, canonical, de-optimize, or internal linking.

CONSTRAINTS
Base yourself only on the data provided. If a case is uncertain, ask for the missing data instead of deciding.

OUTPUT FORMAT
Table: | Cluster (query) | Conflicting URLs | Page to prioritize | Recommended action | Priority |
Then the execution order (from quick win to heavy project).

You are an SEO content strategist. Write a complete brief for content targeting:
- Primary keyword: [KEYWORD]
- Validated intent: [PASTE THE SUMMARY FROM PROMPT 3]
- Market / language: [COUNTRY / LANGUAGE]
- Brand / own angle: [WHAT MAKES US LEGITIMATE — experience, data, client cases]

THE BRIEF MUST CONTAIN
1. Unique editorial angle (why our page deserves to rank vs the existing ones)
2. 3 title proposals (≤ 60 char.) + 1 meta description (≤ 155 char.)
3. Full heading structure (H1 → H2 → H3) covering the intent
4. Entities, sub-topics and semantic terms to integrate mandatorily
5. Questions to address (People Also Ask + questions asked to AI)
6. E-E-A-T signals to include (proof of experience, sources, author, quantified data)
7. 3 to 5 "answer passages": short, standalone, citable sentences that directly answer a question (for both featured snippet AND AI citation)
8. Suggested internal linking + conversion intent / CTA
9. Target length + register

CONSTRAINTS
No keyword stuffing. Every section must have a reason to exist tied to the intent.

OUTPUT FORMAT: structured brief, ready to hand to a writer.

You are a senior SEO writer. From this brief:

[PASTE THE BRIEF FROM PROMPT 5]

MISSION
Produce the detailed OUTLINE of the article (not the writing). For each section (H2/H3):
- the angle and goal of the section
- the precise question it answers
- the entities / figures / examples to place in it
- an "extractable answer": 1 to 2 standalone sentences, sourced if possible, that directly answer the question (placed at the start of the section)
- the relevant rich element (table, list, definition box, visual)

Start with a hook that gives the main answer right from the introduction ("answer-first" model: don't make the reader or the AI wait).

CONSTRAINTS
Progressive reading logic (from what → how → proof). No empty filler sections.

OUTPUT FORMAT
Hierarchical H1/H2/H3 outline with, under each title, the 4 bullets above.

You are an SEO information architect, an expert in topical authority.

PILLAR TOPIC: [BRAND'S CENTRAL THEME]
Activity: [WHAT WE SELL] · Market: [COUNTRY / LANGUAGE]

MISSION
Design a complete semantic map that covers the topic exhaustively in the eyes of Google AND of AI:
1. The PILLAR page (the central hub) and its intent
2. The thematic CLUSTERS (major sub-topics)
3. For each cluster, the satellite articles (intent + target keyword)
4. The key ENTITIES to cover to be recognized as an authoritative source on the topic
5. The INTERNAL LINKING plan: who links to whom, and with which anchor
6. The "coverage gaps" — sub-topics competitors cover that will need to be filled

CONSTRAINTS
Aim for topical completeness, not article quantity. Each page must have a distinct intent (anti-cannibalization, cf. prompt 4).

OUTPUT FORMAT
Tree: Pillar → Clusters → Satellites (with intent + keyword), then internal linking table | Source | Target | Suggested anchor |.

You are an SEO consultant in charge of a "content decay" audit.

EXISTING ARTICLE (paste the text or the URL + content):
[PASTE THE CURRENT CONTENT]

Target keyword: [KEYWORD]
Current performance: [POSITION / CLICKS / TREND — or "declining for X months"]
What ranks today: [PASTE THE SERP SUMMARY FROM PROMPT 3, if available]

MISSION
1. Spot the signs of obsolescence: outdated info/figures/dates, sections that have become false
2. List the sub-topics, entities and questions now expected (vs the current SERP) and absent from the article
3. Identify the weak / too-thin / off-intent sections
4. Detect the internal-linking and "answer passage" opportunities to add (GEO)
5. Assess the effort vs the impact of each improvement

DELIVER
- A PRIORITIZED refresh checklist (Impact × Effort)
- The rewrite of the 2-3 weakest sections
- The new titles/meta if needed

CONSTRAINTS: don't rewrite what already works; don't delete content that ranks without justifying it.

You are a GEO (Generative Engine Optimization) expert. Goal: make this content CITABLE by generative engines (ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude).

CONTENT TO OPTIMIZE:
[PASTE THE PASSAGE / THE ARTICLE]

Target topic / question: [THE QUESTION WE WANT TO BE THE ANSWER TO]

MISSION — apply and show the proven citability levers:
1. DIRECT ANSWER up front: a clear, standalone, 1-3 sentence statement that answers the question with no prior context.
2. STATISTICS & quantified DATA (engines cite content with figures more often).
3. SOURCES & CITATIONS: attribute the facts ("according to [source]…"), add verifiable references.
4. Crisp DEFINITIONS of the key entities/concepts ("X is…" format).
5. CITABLE PHRASING: short, factual sentences, extractable out of context.
6. STRUCTURE: question-form headings, lists, comparison tables, callout boxes.
7. AUTHORITY SIGNALS: author, expertise, freshness (date), consistency with the field's consensus.

DELIVER
- The rewritten version, optimized for citation
- A callout box "Why this version will be cited" (levers activated)
- 3 "snippet-ready" rephrasings of the main answer

ABSOLUTE CONSTRAINT: zero invention. If a statistic or source is missing, write [TO BE SOURCED] — never fabricate a figure.

You are a multi-engine GEO strategist. For the content / query below, give a DIFFERENTIATED optimization plan per generative engine.

Target content or query: [PASTE THE CONTENT OR THE QUESTION]
Brand: [BRAND] · Market: [COUNTRY / LANGUAGE]

For EACH of the engines, state what it rewards and 3-4 concrete actions:
- GOOGLE AI OVERVIEWS / AI MODE: breakdown into sub-questions (query fan-out), indexable passages, entities + structured data, presence in the organic top, freshness.
- CHATGPT (search + training data): notable, authoritative sources, repeated and consistent brand mentions across the web, clear factual phrasing, well-established pages.
- PERPLEXITY: real-time citations, freshness, multiple corroborating sources, structured content that's easy to source.
- GEMINI: Google ecosystem (entities, schema, consistency), recognized and linked content.
- CLAUDE: reliable and verifiable sources, structured content, factual tone.

DELIVER
- A table | Engine | What it rewards | My 3 priority actions |
- The COMMON points (to do once, valid everywhere)
- The high-leverage SPECIFIC actions per engine

CONSTRAINT: distinguish what is content (on-page) from what is off-site authority (mentions, third-party sources).

You are an AI-visibility consultant. Design an AUDIT PROTOCOL to measure the brand's presence in generative answers.

Brand: [BRAND]
Sector / category: [CATEGORY]
Competitors: [2 TO 5 COMPETITORS]
Market / language: [COUNTRY / LANGUAGE]

MISSION
1. Generate 15 to 25 realistic TEST PROMPTS that customers would ask an AI in this category (research, comparison, recommendation questions, "best X for Y").
2. Define the MEASUREMENT GRID to fill for each prompt, on each engine (ChatGPT, Perplexity, Gemini, Google AI Overviews):
   - is the brand cited? (yes/no)
   - at what position / in what order vs competitors?
   - with what sentiment / what angle?
   - which SOURCES does the AI cite (which sites let it answer)?
3. Define the "AI share of voice" indicator for the brand vs competitors.
4. From the cited sources, deduce WHERE you need to be present/mentioned to enter the answers.

DELIVER
- The list of test prompts
- The blank tracking table to fill
- The share-of-voice calculation method
- An action plan to improve presence (content + third-party sources to target)

CONSTRAINT: it's a reproducible protocol (to re-run every month), not a one-off audit.

You are an expert in technical SEO + structured data (schema.org).

Page / topic: [TOPIC OR URL]
Target keyword: [KEYWORD]
Market / language: [COUNTRY / LANGUAGE]
Brand info for the schema: [NAME, URL, LOGO, SOCIALS, ORGANIZATION TYPE]

MISSION — PART 1: FAQ
Generate 6 to 10 questions actually asked about this topic (People Also Ask style + questions asked to AI), with answers:
- short, standalone, factual (40-60 words), usable as a featured snippet and citable by an AI
- no keyword stuffing, in natural language

PART 2: STRUCTURED DATA
Produce the VALID JSON-LD, ready to paste, wrapped in a <script type="application/ld+json"> tag:
- FAQPage (from the Q/A above)
- Article (or the most suitable class: Product, Service, LocalBusiness, MedicalOrganization…)
- Organization (with sameAs to the socials)
- BreadcrumbList if relevant

CONSTRAINTS
- Schema 100% schema.org-compliant, no invented field; put [TO COMPLETE] wherever I'm missing data.
- Warn if the page isn't eligible for the FAQ rich result (Google has restricted the display to government/health sites on certain queries).

OUTPUT FORMAT: 1) FAQ block (Q/A) — 2) ready-to-paste JSON-LD block(s).