seo-content-homepage
Skill18 starsChanged 11 days agoMITView on GitHub
Runs spec, audit, competitor benchmark and build for the homepage and the flagship offer page (sales page) of an expert, coach, trainer, agency, consultant or SaaS, so they rank, get cited by AI engines and convert. Phase 1 is the spec: HOME and OFFER requirement tables (the 16 homepage blocks, the sales page sequence, semantic bolding, a number in every block, an objection FAQ, a visible price, dated ratings, checkable credentials, real social profiles, honest claims and regulated-topic notices) on top of the common page spec. Phase 2 audits the live home, offer and about pages with seo_audit.py and section_audit.py. Phase 3 benchmarks 3-5 competitors with page_benchmark.py and picks the information gain. Phase 4 writes the copy block by block with Organization, WebSite, Person and Course or Service JSON-LD, then re-runs the scripts as the acceptance test. Use when a homepage or sales page is thin, has no bold, no numbers, no proof, only a generic FAQ, or is being built from scratch.
In this repository27 other files it gives its agents.
- AGENTS.mdAGENTS.md
- geo-trackingSkill
- geo-visibilitySkill
- seo-ai-site-buildersSkill
- seo-audit-pdfSkill
- seo-backlinksSkill
- seo-competitor-analysisSkill
- seo-content-blogSkill
- seo-content-collection-pageSkill
- seo-content-comparison-hubSkill
- seo-content-comparison-pageSkill
- seo-content-data-studySkill
- seo-content-product-pageSkill
- seo-content-service-pageSkill
- seo-geo-auditSkill
- seo-internal-linkingSkill
- seo-keyword-researchSkill
- seo-localSkill
- seo-page-sectionsSkill
- seo-schema-markupSkill
- seo-sitemapSkill
- seo-technicalSkill
- seo-traffic-dropSkill
- social-amplificationSkill
- youtube-descriptionSkill
- youtube-thumbnailSkill
- youtube-titleSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 41
Ask for, or pull from the vault, before writing a line.
- Distinctive: line 127
Run both collectors on the three pages (raw HTML, no JavaScript):
- Very specific: line 136
Use --type pricing for the sales page and for a separate pricing page (its wireframe is in transverse-pages.md).
- Distinctive: line 165
Prefer the brands that appear in two or more of these sources.
- Very specific: line 186
Read references/block-library.md now: per block minimum spec, the typical objection set, the offer sequence and the about page.
- Distinctive: line 251
No high or critical finding.
- Distinctive: line 259
Every C, HOME, OFFER and ABOUT row is Pass, or a remaining Fail is a written owner decision.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Homepage and offer page that rank, get cited and sell
The homepage is the strongest page of a small site: most backlinks land there, every branded search lands there, and AI assistants answering "who is {brand}" or "best {category} coach" read it first. The flagship offer page is where the money is made. On expert and coaching sites both are usually a stack of slogans, three icons, a video and a generic FAQ ("What is poker?"): 800 words, no bold, no numbers, no structured data. That page cannot rank on the category term, gives an AI engine nothing to quote and gives a buyer no reason to trust.
This skill fixes that in four phases (spec, audit, benchmark, build, see docs/page-skill-standard.md) with a measurable finish line.
Company knowledge first (Obsidian)
If the working environment contains an Obsidian vault or any local knowledge base (a folder of .md notes, often with a .obsidian directory), read the relevant notes before writing: offer facts, real prices, real client results, founder story and credentials, positioning, competitors, social profile URLs, and the SEO action log. Every block of a homepage asserts facts about the business, so invented numbers are the main failure mode. At the end of the session, append what was published to the vault's SEO action log.
When to use
- A homepage or sales page is thin, slogan-only, or has only a generic FAQ.
- An audit (
seo_audit.py) reportsno_bold,thin_content,schema_none,schema_entity,home_no_money_links,social_placeholderor a low placement coverage on the homepage. - A site is being built from scratch, or generated by an agent, and the block list must be decided first.
- The owner asks "what are my competitors doing that I am not".
| Adjacent need | Skill |
| Service pages and city pages | seo-content-service-page |
| Articles, the blog hub and article templates | seo-content-blog |
| Block-by-block gap audit of any other page type; the pricing, about and contact wireframes | seo-page-sections (references/transverse-pages.md) |
| JSON-LD fields and templates (Course template included) | seo-schema-markup |
| Choosing the category term and its variants | seo-keyword-research |
| "{brand} vs {competitor}" and alternatives pages | seo-content-comparison-page |
| Which brands AI assistants name, citation tracking | geo-visibility, geo-tracking |
Phase 1. The spec
1.1 Facts the spec needs
Ask for, or pull from the vault, before writing a line. Missing facts become visible placeholders {to confirm}, never inventions.
| Fact | Used in |
| Category term and its search variants (what buyers type: "coaching poker", "formation poker", "agence SEO Lyon") | Title, H1, first 100 words, H2s (HOME-20) |
| The outcome promised, with a number and a timeframe | Hero sub-line, results block (HOME-02, HOME-07) |
| Proof numbers: clients served, years, results, ratings with their platform and count | Proof strip, results, founder block (HOME-03, HOME-22, HOME-25) |
| 3-6 client results with first name, starting point, result, timeframe; video testimonials | Results and testimonials (HOME-07, HOME-08) |
| Founder credentials that can be checked (years, track record, press, certifications, public profiles) | Founder block, Person schema (HOME-09, HOME-26) |
| The offer: format, contents, time commitment, price or price range, guarantee, process to join | Offer cards, sales page (HOME-06, OFFER-05 to OFFER-14) |
| Real social profile URLs (LinkedIn, YouTube, Instagram, X, TikTok, community) | Footer, founder block, sameAs (HOME-27) |
| The 8-12 objections prospects actually raise (sales calls, emails, DMs, comments) | FAQ (HOME-15, OFFER-16) |
1.2 Homepage requirements (HOME)
Common requirements C-01 to C-27: skills/seo-geo-audit/references/common-page-spec.md. They apply to all three pages and are not repeated here. Thresholds are field heuristics from 115+ agency audits unless marked measured; "14 sites" counts come from the October 2026 coaching benchmark in references/block-library.md section 5. Block details (minimum spec per block) live in references/block-library.md section 1, same numbering.
| ID | Requirement | Threshold | Verified by | Why |
| HOME-01 | Header | Logo, 4-6 menu items (offers, results, about, blog, contact), one CTA button in a contrasting color, sticky | section_audit cta; manual | The CTA stays visible at every scroll position |
| HOME-02 | Hero | H1 = category term + outcome, never a slogan; sub-line with one proof number; primary CTA (main conversion) and secondary CTA (low commitment: free training, guide, quiz); real founder or product image, eager-loaded | h1_missing, kw_not_in_intro, lcp_lazy; manual for the outcome | The first screen answers what, for whom, why trust, what next |
| HOME-03 | Proof strip | Rating with platform and count, client count, years, press or partner logos | section_audit rating_summary, social_proof_logos, trust_badges | Borrowed authority before any claim |
| HOME-04 | Problem | 3-6 pains in the buyer's own words (from calls, comments, reviews), each with what changes, at least one with a number | manual | Recognition keeps the right reader scrolling |
| HOME-05 | Named method | A name and 3-5 pillars, each with one line and one number | section_audit process_block; manual | What competitors cannot copy and what AI engines quote as an entity |
| HOME-06 | Offer routing | One card per offer: who it is for, format, price or "from", body link to its page | home_no_money_links; section_audit internal_links, price_in_text | Routes the homepage authority to the money pages |
| HOME-07 | Results with numbers | 3-6 results: first name, starting point, result, timeframe; screenshots or graphs; links to full case studies | no_stats; manual | Specific outcomes are the strongest conversion and citation material; numbers-based proof is strong on only 4 of 14 sites (measured) |
| HOME-08 | Testimonials | Video (YouTube embed with a text summary under each) plus written ones with real first name, photo, situation | section_audit reviews, video | Video proves the person exists; the summary makes it readable by crawlers |
| HOME-09 | Founder block | Real photo, 80-150 words, credentials with numbers, link to the about page, real social profiles | section_audit author_block; manual | E-E-A-T: who stands behind the promise |
| HOME-10 | Comparison table | This offer versus the usual alternatives (free content, other programs, doing it alone), 5-8 criteria, an HTML table | section_audit comparison_table; home_no_structure | Tables are the most extracted format in AI answers and the most direct answer to "why you"; 8 of 14 sites compare (measured) |
| HOME-11 | For / not for | Two short lists | home_no_structure; manual | Qualification raises conversion and lowers refunds |
| HOME-12 | Guarantee | Exact terms (duration, condition), only when one exists | section_audit guarantee | Risk reversal; vague guarantees read as traps; 3 of 14 sites, all English (measured) |
| HOME-13 | Lead magnet | Free training, guide, checklist, quiz or tool with an email form | section_audit form; manual | Captures the 95% not ready to buy today (field heuristic); 11 of 14 sites have one (measured) |
| HOME-14 | Latest articles | 3 cards with cover image, title, excerpt, date | section_audit freshness_signal; manual | Shows the site is alive and links the blog from the strongest page |
| HOME-15 | Objection FAQ | 8-12 real objections (price, time, level, difference from free content, results timeline, refund, support, how to join), 2-4 sentence answers, <details> with the text in the HTML; no definitions | section_audit faq; manual for the questions | Objections answered in writing replace a sales call; question-answer pairs are extracted by AI engines |
| HOME-16 | Descriptive block | 300+ words of prose on the category term, H2 phrased as the question buyers ask, bold on key facts | thin_content; manual | Gives retrieval and ranking real text when the rest of the page is short blocks |
| HOME-17 | Final CTA | Promise, one proof, one button | section_audit cta | The last conversion surface |
| HOME-18 | Footer | Real social profiles, contact, legal pages, disclaimers, sitemap-level links | social_placeholder, social_missing; section_audit contact_details | Entity consistency and trust; never placeholder icons |
| HOME-19 | Length | 1500-1800 words of real text, never under the competitor median (C-08 floor for this page type) | thin_content; page_benchmark.py words | A page of headlines gives ranking and retrieval nothing to work with |
| HOME-20 | Category term is the target keyword of C-01, C-03 and C-10 | Title "{what the business does} for {audience} | {brand}", never "{Brand} | Home"; term in title, H1, first 100 words, 2+ H2s, meta description, one image alt; placement 60%+ | kw_placement_low, kw_not_in_intro | The H1 is the second strongest relevance signal after the title; the homepage title is the highest-authority title on the site |
| HOME-21 | Semantic bolding (C-09 with the category term) | Category term bold once in the first paragraph; then one bold phrase every 100-150 words on a benefit, a number or a key fact; never whole sentences, never decoration | no_bold, kw_not_bold | Cheapest visible relevance and scannability signal; skimmers read the bold and the H2s only |
| HOME-22 | Numbers in every block | Each block carries at least one figure: results, clients, years, hours, price, timeframe | no_stats; citability statistics; manual per block | Statistics raise generative visibility (Princeton GEO study, about +32%, measured); specific claims convert better than adjectives |
| HOME-23 | Proof attribution | Each number says where it comes from: client name, platform, date, screenshot | manual (extends C-14) | Unattributed numbers read as marketing to buyers and quality systems |
| HOME-24 | Visible price | Price, range or "from" on the offer cards and the offer page, with instalments or an annual option when they exist; if the price is given only after a call, say so and why | section_audit price_in_text, pricing_block | 12 of 14 coaching sites show a price, 5 show payment options (measured); hiding it sends buyers and assistants to the ones that do |
| HOME-25 | Dated ratings | Rating + platform + review count + date read ("4.5/5 Trustpilot, 1253 avis, relevé le 9 août 2026") | section_audit rating_summary; manual for the date | A dated, counted rating is checkable, a bare star row is not; only 5 of 14 sites do it (measured) |
| HOME-26 | Checkable credentials | Founder linked to third-party profiles that confirm the claims (the field's public results databases, press, Wikipedia or Wikidata when they exist); the same URLs in Person sameAs | manual; schema_fields | Claims an assistant can cross-check become entity facts; claims it cannot are ignored; 4 of 14 sites do it (measured) |
| HOME-27 | Real social profiles on both entities | C-20 applied to the footer and the founder block; Organization sameAs and Person sameAs list the same real URLs | social_placeholder, social_missing, schema_fields | Entity consistency between the site, the profiles and the graph |
| HOME-28 | Honest claims (C-24 applied) | Results stated as observed outcomes of named clients, never as a guarantee; "n°1", "le meilleur", "leader" only with a cited source next to them; countdowns and limited places only when real | claim_no_disclaimer, superlative_claim; manual for urgency | Unbacked superlatives lower trust; fake urgency is a deceptive commercial practice under EU consumer law |
| HOME-29 | Regulated-topic notices | Income or results disclaimer next to the claim and in the footer; gambling: 18+, responsible-gambling mention, national helpline in the site-wide footer, a responsible-play page; editorial transparency pages linked from the footer | claim_no_disclaimer, gambling_notice_missing; manual | Your Money or Your Life topics; ARPP and FTC rules on testimonials with results; 7 of 14 poker coaching sites show the notice (measured). Detail: "Regulated and money topics" below |
| HOME-30 | Body links | Every offer page, the about page and 2-3 pillar articles linked from the body with descriptive anchors, not only the menu (C-16 applied) | home_no_money_links, generic_anchor; section_audit internal_links | The homepage's authority flows only through body links with descriptive anchors |
| HOME-31 | Homepage schema | Organization (name, url, logo, sameAs to every real profile, founder) + WebSite + the founder as Person with sameAs, one @graph; nothing marked up that the page does not show | schema_none, schema_entity, schema_fields, jsonld_invalid | Entity graph for AI engines; 11 of 14 sites ship a JSON-LD graph (measured), the client without it is in the bottom 3 |
| HOME-32 | Contact reachable | Contact details or a contact route reachable from the homepage | section_audit contact_details | A buyer ready to act must not hunt |
| HOME-33 | Shared competitor blocks | Every block present on 3+ benchmarked competitors (page_benchmark.py: half or more) is present | page_benchmark.py block gaps | The SERP shows which page shape Google already rewards; that list is the minimum, not the ceiling |
1.3 Flagship offer page requirements (OFFER)
The sales sequence, in this order. HOME-20 to HOME-30 apply to this page too (category or offer name as keyword, bolding, numbers, attribution, price, ratings, credentials, profiles, honest claims, notices, body links). Block details: references/block-library.md section 2, same numbering.
| ID | Requirement | Threshold | Verified by | Why |
| OFFER-01 | Hero | H1 = offer name + outcome; sub-line with the promise and timeframe; video sales letter or founder video (YouTube embed with chapters or a text summary); primary CTA | h1_missing; section_audit video, cta | The first screen sells the outcome, not the format |
| OFFER-02 | Proof strip | Rating, client count, result headline | section_audit rating_summary, social_proof_logos | Trust before the pitch |
| OFFER-03 | Problem and the cost of staying put | Buyer's words, with numbers | manual | Makes the price read as smaller than the cost of inaction |
| OFFER-04 | The method | Named, 3-5 pillars, each linked to what the client does with it | section_audit process_block; manual | The method is the product's identity |
| OFFER-05 | Program contents | Modules or phases with titles and one-line outcomes; number of lessons, hours, live sessions | section_audit any_table; manual | 7 of 14 sites list a curriculum (measured); counts make the offer concrete |
| OFFER-06 | Format and workload | Duration, hours per week, live versus recorded, access length, community, support channels and response times | manual | Answers "will this fit my week"; 12 of 14 sites show their community (measured) |
| OFFER-07 | Results | Case studies with before, after, timeframe, screenshots or graphs | no_stats; manual | The strongest proof on the page |
| OFFER-08 | Testimonials | Video and text, named | section_audit reviews, video | Named voices beat anonymous quotes |
| OFFER-09 | The coach or team | Photo, credentials, link to the about page | section_audit author_block | Who delivers the promise |
| OFFER-10 | Comparison table | Versus the alternatives, 5-8 criteria, real costs of each alternative | section_audit comparison_table | Answers "why this and not a private coach or free content" |
| OFFER-11 | Price and payment | Price as HTML text (or "from"), payment plans, what is included; if only after a call, say so and why | section_audit price_in_text, pricing_block | A price in an image or behind JavaScript is invisible to AI crawlers (C-21) |
| OFFER-12 | Guarantee | Exact terms, only when one exists | section_audit guarantee | A real guarantee is a differentiator in French markets |
| OFFER-13 | Bonuses | Only real ones, each with its standalone value | manual | Inflated bonuses read as a trap |
| OFFER-14 | Process to join | 3-4 numbered steps (apply, call, onboarding) with durations; booking link; what happens on the call | section_audit process_block, form | Removes the "what happens next" doubt; 5 of 14 sites run a call or application (measured) |
| OFFER-15 | For / not for | Two lists | manual | Qualification lowers refunds |
| OFFER-16 | Objection FAQ | 10-15 real objections | section_audit faq | The written sales call |
| OFFER-17 | Final CTA | Promise, proof, button | section_audit cta | The last conversion surface |
| OFFER-18 | Length | 2000+ words, never under the competitor median | thin_content; page_benchmark.py words | A sales page must answer every objection in text |
| OFFER-19 | Offer schema | Course (curriculum program: name, description, provider, offers, hasCourseInstance), Service (one-to-one coaching) or Product with Offer (digital product); BreadcrumbList; Organization referenced by @id; schema price equals the visible price; nothing not shown on the page | schema_none, schema_fields, jsonld_invalid; manual for the match | A price, rating or session marked up but not visible is a spam policy risk |
| OFFER-20 | Shared competitor blocks | Every block half or more of the benchmarked offer pages have is present | page_benchmark.py block gaps | Same as HOME-33 |
1.4 About or founder page (ABOUT)
The entity anchor and, on a one-expert site, the author page every article links to. The wireframe is owned by seo-page-sections (references/transverse-pages.md, About page); the founder variant is in references/block-library.md section 3. Blocks: H1 full name + role; answer-first lead (what, since when, for whom, one proof number); facts block (years, results with sources, clients, locations, certifications, press); honest dated story; proof linked to third-party sources; real social profiles; the author's articles and a CTA to the offer; ProfilePage with mainEntity Person (same @id as article bylines) and sameAs. Consistency rule: name, one-line description, founding year and location identical on the site, the profiles and the schema.
Phase 2. Audit the existing page
Run both collectors on the three pages (raw HTML, no JavaScript):
python3 skills/seo-geo-audit/scripts/seo_audit.py https://example.com/ /offer/ /about/
python3 skills/seo-page-sections/scripts/section_audit.py --type homepage https://example.com/
python3 skills/seo-page-sections/scripts/section_audit.py --type pricing https://example.com/offer/
python3 skills/seo-page-sections/scripts/section_audit.py --type about https://example.com/about/Use --type pricing for the sales page and for a separate pricing page (its wireframe is in transverse-pages.md). Map every finding to a row and mark each C, HOME, OFFER and ABOUT row Pass, Fail or Not verifiable:
| Finding or block | Rows |
| no_bold, kw_not_bold | C-09, HOME-21 |
| kw_placement_low, kw_not_in_intro, title_short, title_long | C-01, C-10, HOME-20 |
| thin_content | C-08, HOME-16, HOME-19, OFFER-18 |
| no_stats | C-14, HOME-07, HOME-22, OFFER-07 |
| home_no_money_links, generic_anchor | C-16, HOME-06, HOME-30 |
| home_no_structure | HOME-10, HOME-11 |
| social_placeholder, social_missing | C-20, HOME-18, HOME-27 |
| schema_none, schema_entity, schema_fields, jsonld_invalid | C-18, HOME-31, OFFER-19, ABOUT |
| claim_no_disclaimer, gambling_notice_missing, superlative_claim | C-24, HOME-28, HOME-29 |
section_audit cta, rating_summary, social_proof_logos, trust_badges | HOME-01, HOME-03, HOME-17, HOME-25, OFFER-01, OFFER-02, OFFER-17 |
section_audit process_block, reviews, video, author_block | HOME-05, HOME-08, HOME-09, OFFER-04, OFFER-08, OFFER-09, OFFER-14 |
section_audit comparison_table, guarantee, form, faq | HOME-10, HOME-12, HOME-13, HOME-15, OFFER-10, OFFER-12, OFFER-16 |
section_audit price_in_text, pricing_block | HOME-06, HOME-24, OFFER-11 |
section_audit contact_details, freshness_signal, breadcrumb | HOME-32, HOME-14, C-19 |
Rows marked "manual" are judged by reading the page (is the H1 an outcome or a slogan, are the FAQ questions objections or definitions, does every block carry a number). "Not verifiable" when the block is rendered by JavaScript (absence in raw HTML is itself a C-21 finding, check with render_check.py) or when the fact needs the owner (a rating date, a credential). When no page exists yet, skip this phase: Phase 3 sets the bar.
Phase 3. Benchmark the competitors
Pick 3-5 competitors, homepage against homepage and offer page against offer page:
- The category term SERP: the businesses ranking organically for what buyers type ("coaching poker"), not directories, marketplaces or Wikipedia.
- The "best {category}" query ("meilleur coaching poker", "best SEO agency Lyon"): the brands the lists and roundups name.
- The brands AI assistants name: ask ChatGPT, Perplexity, Gemini and Google AI Overviews "best {category} for {audience}" and note who is recommended (geo-visibility, geo-tracking).
Prefer the brands that appear in two or more of these sources. Then:
python3 skills/seo-geo-audit/scripts/page_benchmark.py --type homepage https://client.com/ https://rival1.com/ https://rival2.com/ https://rival3.com/
python3 skills/seo-geo-audit/scripts/page_benchmark.py --type pricing https://client.com/offer/ https://rival1.com/programme/ https://rival2.com/coaching/ https://rival3.com/formation/What to read in the output:
- The metric table: words, bold, H2, question headings, tables, statistics, placement, money links, social profiles, videos, schema types per site.
- "BLOCKS half or more of the competitors have and the client does not": the minimum for HOME-33 and OFFER-20, not the ceiling.
- "SCHEMA TYPES competitors use and the client does not": feeds HOME-31 and OFFER-19.
- "CLIENT BELOW THE COMPETITOR MEDIAN": the C and HOME rows behind those metrics move up in priority; the median word count sets HOME-19 and OFFER-18.
- What the script cannot measure, read by hand on each competitor: proof formats (numbers, video, screenshots, ratings), pricing visibility, lead magnet, free tools or resources, community, named method, guarantee.
references/block-library.mdsection 5 gives how often each appeared on 14 coaching and training sites, to judge what is table stakes and what is rare.
Information gain: from the blocks few or none of the competitors ship, choose 1-3 the client can back with real facts and write them down. Typical candidates: original numbers (named student results, a results graph), a comparison table of the real costs of each alternative, a named method, a free tool, a money-back guarantee in a French market, a coach linked to a checkable third-party profile, a dated rating with its count. The conversion blocks in references/block-library.md section 4 list when each is worth adding. Why: the SERP shows which page shape Google already rewards for this query, and the competitors' strongest proof is the bar the client's proof must clear.
Phase 4. Build
4.1 Block order
Read references/block-library.md now: per block minimum spec, the typical objection set, the offer sequence and the about page. Keep every block the benchmark shows on 3+ competitors, then add the information gain chosen in Phase 3.
Default homepage order:
- Header with the primary CTA button (HOME-01)
- Hero: H1 = category + outcome, sub-line with one proof number, primary and secondary CTA, proof strip: rating, client count, logos or press (HOME-02, HOME-03)
- Problem block in the buyer's words (HOME-04)
- The method, named, in 3-5 pillars (HOME-05)
- Offer routing: one card per offer or service, each linked to its page (HOME-06)
- Results with numbers (before, after, timeframe), case study links (HOME-07)
- Testimonials: video plus text, real first name, photo, situation (HOME-08)
- Founder block: photo, credentials with numbers, link to the about page, social profiles (HOME-09)
- Comparison table: this offer versus the usual alternatives (HOME-10)
- Who it is for / not for (HOME-11)
- Guarantee or risk reversal, when there is one (HOME-12)
- Lead magnet or free resource (HOME-13)
- Latest articles: 3 cards with cover images (HOME-14)
- FAQ: 8-12 real objections, not definitions (HOME-15)
- Descriptive block of 300+ words on the category term (HOME-16)
- Final CTA, then a footer with real social profiles, contact, legal links (HOME-17, HOME-18)
The offer page follows OFFER-01 to OFFER-17 in order: outcome, proof, problem, method, program contents, format and workload, results, testimonials, coach, comparison, price and payment, guarantee, bonuses, process to join, for / not for, FAQ, final CTA.
4.2 Copy rules
Field heuristics from 115+ agency audits unless marked measured.
| Rule | Detail | Rows |
| Length | Homepage 1500-1800 words of real text, offer page 2000+; never under the competitor median | HOME-19, OFFER-18 |
| H1 | Category term + outcome, not a slogan ("Coaching poker pour devenir rentable en cash game", not "Libère ton potentiel") | HOME-02, C-03 |
| Placement | Category term in title, H1, first 100 words, at least two H2s, meta description, one image alt, and in bold once in the intro; target 60%+ coverage of the placement matrix | HOME-20, C-10 |
| Semantic bolding | Category term bolded once in the first paragraph; one bold phrase every 100-150 words on a benefit, a number or a key fact; never whole sentences, never decoration | HOME-21, C-09 |
| Numbers | Every block carries at least one figure | HOME-22 |
| Proof attribution | Each number says where it comes from (client name, platform, date, screenshot) | HOME-23, C-14 |
| Density | No term above 2.5% of the text, brand names aside | C-11 |
| Question H2s | At least 3 H2s phrased as the questions buyers ask ("Combien de temps faut-il pour devenir rentable ?"); they map to People Also Ask and to AI sub-queries | C-13 |
| Chunk autonomy | Each block understandable alone, 40-80 word paragraphs; AI engines lift isolated passages | C-12 |
| Internal links | Every offer page, the about page and 2-3 pillar articles linked from the body | HOME-30, C-16 |
| Punctuation | Zero em dashes and en dashes, the most recognizable AI-writing tell | C-25 |
| Price | Price, range or "from", instalments or annual option; if only after a call, say so and why | HOME-24, OFFER-11 |
| Ratings | Rating + platform + count + date read | HOME-25 |
| Checkable credentials | Third-party profiles linked from the founder block and listed in sameAs | HOME-26 |
| Superlatives | "n°1", "le meilleur", "leader" only with a cited source next to them, otherwise remove | HOME-28 |
| Urgency | Countdowns and limited places only when the deadline or the limit is real | HOME-28 |
4.3 Metadata patterns
Homepage title: {what the business does} for {audience} | {brand} (50-60 chars)
Homepage meta: {outcome}. {proof number}. {call to action}. (120-160 chars)
Homepage H1: {category term} + {outcome}
Offer page title: {offer name}, {outcome} | {brand} (50-60 chars)
Offer page H1: {offer name} + {outcome}
About H1: {full name}, {role} ({brand})4.4 Structured data
Homepage: Organization (name, url, logo, sameAs to every real profile, founder) plus WebSite, and the founder as a Person with sameAs (HOME-31). Offer page: Course when it is a program with a curriculum, Service when it is one-to-one coaching, Product with Offer for a digital product; BreadcrumbList on both offer and about pages (OFFER-19, C-19). About page: ProfilePage with the Person. Templates and field rules: seo-schema-markup (Course template in its jsonld-templates reference). Mark up only what the page shows: a price, a rating or a session that is not visible on the page is a spam policy risk.
4.5 Acceptance test
Re-run the Phase 2 commands and the Phase 3 benchmark on the live or staged pages. The finish line:
- No high or critical finding.
- BOLD count above zero with the category term among the samples.
- PLACEMENT coverage at 60% or more.
- No
schema_none,schema_entityorschema_fields. - SOCIAL shows real profiles and no placeholders.
home_no_money_linksabsent.- Word count at or above the competitor median.
section_audit.pyblock list compared with the benchmark: every block present on 3+ competitors is now present (page_benchmark.py block gap list empty).- Every C, HOME, OFFER and ABOUT row is Pass, or a remaining Fail is a written owner decision.
Regulated and money topics (HOME-28, HOME-29)
Gambling, trading, income, health, weight loss: state results as observed outcomes of named clients, never as a guarantee; add an income or results disclaimer next to the claim and in the footer ("aucun résultat individuel n'est garanti"); for gambling, add the age limit (18+), a responsible-gambling mention and the national helpline in the site-wide footer (in France: Joueurs Info Service, 09 74 75 13 13), plus a responsible-play page. The audit flags claim_no_disclaimer and gambling_notice_missing. Editorial transparency pages (editorial policy, methodology, how results are measured) linked from the footer strengthen the same trust layer. Google rates these topics "Your Money or Your Life": trust signals are judged more strictly, and advertising rules (ARPP in France, FTC endorsement guides in the US) apply to testimonials with results.
What the coaching benchmark taught (HOME-24 to HOME-31)
The full table is in references/block-library.md section 5 (14 poker coaching and training sites, October 2026, measured). What it changes for any coaching or training site: a JSON-LD graph, a visible price, a community and a lead magnet are table stakes (11-12 of 14); numbers-based proof, a coach linked to checkable third-party records, a dated rating with its count and a money-back guarantee are rare (3-5 of 14) and are the cheapest information gain.
Deliverable
## 1. Spec scorecard
| ID | Requirement | Status (Pass / Fail / Not verifiable) | Evidence |
(C-01 to C-27, HOME-01 to HOME-33, OFFER-01 to OFFER-20, ABOUT blocks)
## 2. Audit findings
(seo_audit.py SEO and GEO scores per page and high or medium findings; section_audit.py blocks for home, offer and about; or "new page")
## 3. Competitor benchmark
(how the competitors were picked; page_benchmark.py table for the homepage and the offer page; blocks and schema types the client lacks; metrics below the median; proof formats, pricing, lead magnet, community read by hand; the information gain chosen)
## 4. Build
- Title (30-60 chars), meta description (120-160 chars), H1, category term and variants
- Page copy block by block, in order ([1. HERO] ...), bolding visible in markdown, placeholders {to confirm} for missing facts
- JSON-LD: Organization + WebSite + Person graph for the homepage; Course, Service or Product with Offer plus BreadcrumbList for the offer page; ProfilePage for the about page
- Internal links added (offer pages, about page, pillar articles)
## 5. Acceptance
(scripts re-run: findings before and after; remaining findings, each either fixed or a written owner decision)Common mistakes
| Mistake | Fix | Row |
| Generic FAQ of definitions ("What is poker?") | FAQ of the real objections: price, time, level required, difference from free content, results timeline, refund | HOME-15, OFFER-16 |
| Slogan H1 | Category term + outcome | HOME-02 |
| Zero bold | Category term bolded in the intro, one bold phrase every 100-150 words on facts and benefits | HOME-21 |
| Testimonials without numbers or names | First name, starting point, result, timeframe, photo or video | HOME-07, HOME-08 |
| Founder block with no credential | Years, verifiable results, profiles, press; link to the about page | HOME-09, HOME-26 |
Footer social icons pointing to instagram.com/ | Real profile URLs or no icon; the same URLs in sameAs | HOME-27 |
| No price anywhere | A price, a range or "from"; hiding it entirely sends buyers and assistants to a competitor who shows one | HOME-24, OFFER-11 |
| Offer page with no structured data | Course, Service or Product with Offer, plus the Organization reference | OFFER-19 |
| Blog invisible from the homepage | 3 latest articles with cover images | HOME-14 |
| Income promises without disclaimer | Observed results of named clients plus a disclaimer; age and responsible-play mentions on gambling topics | HOME-28, HOME-29 |
| Benchmarking against directories or Wikipedia | Compare with businesses of the same page type that rank or that assistants name | Phase 3 |
Sources
- Princeton GEO study, "GEO: Generative Engine Optimization" (KDD 2024): https://arxiv.org/abs/2311.09735
- Google, Search Quality Rater Guidelines (YMYL and E-E-A-T): https://static.googleusercontent.com/media/guidelines.raterhub.com/en//searchqualityevaluatorguidelines.pdf
- Google, Course info structured data: https://developers.google.com/search/docs/appearance/structured-data/course-info
- Google, Organization structured data: https://developers.google.com/search/docs/appearance/structured-data/organization
- FTC, Endorsement Guides: https://www.ftc.gov/legal-library/browse/rules/guides-concerning-use-endorsements-testimonials-advertising
- Coaching benchmark (14 sites, October 2026):
references/block-library.mdsection 5, measured. - Lengths, bold cadence, block order: field heuristics from 115+ agency audits, not Google statements.
- Internal:
docs/page-skill-standard.md,skills/seo-geo-audit/references/common-page-spec.md,skills/seo-geo-audit/references/audit-checklist.mdsection 15,skills/seo-page-sections/references/transverse-pages.md.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

