aeorank 3.0.3 → 3.1.0

package/README.md

@@ -1,6 +1,6 @@
 # AEORank
 
-Score any website for AI engine visibility across 34 criteria in a 5-pillar framework. Pure HTTP + regex - zero API keys, under 10 seconds.
+Score any website for AI engine visibility across 36 criteria in a 5-pillar framework. Pure HTTP + regex - zero API keys, under 10 seconds.
 
 [![npm version](https://img.shields.io/npm/v/aeorank.svg)](https://www.npmjs.com/package/aeorank)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
@@ -35,7 +35,7 @@ import { audit } from 'aeorank';
 
 const result = await audit('example.com');
 console.log(result.overallScore);  // 0-100
-console.log(result.scorecard);     // 34 criteria with scores, pillars, weights
+console.log(result.scorecard);     // 36 criteria with scores, pillars, weights
 console.log(result.pillarScores);  // { answerReadiness, contentStructure, ... }
 console.log(result.topFixes);      // Top 3 highest-impact fixes
 console.log(result.opportunities); // Prioritized improvements
@@ -43,7 +43,7 @@ console.log(result.opportunities); // Prioritized improvements
 
 ## What It Checks
 
-AEORank evaluates 34 criteria that determine how AI engines (ChatGPT, Claude, Perplexity, Google AI Overviews) discover, parse, and cite your content. Criteria are organized into five pillars:
+AEORank evaluates 36 criteria that determine how AI engines (ChatGPT, Claude, Perplexity, Google AI Overviews) discover, parse, and cite your content. Criteria are organized into five pillars:
 
 ### 5-Pillar Framework
 
@@ -55,7 +55,9 @@ AEORank evaluates 34 criteria that determine how AI engines (ChatGPT, Claude, Pe
 | Original Data & Expert Analysis | 10% | Proprietary research, case studies, unique data points |
 | Content Depth | 7% | Article length, heading structure, deep vs thin pages |
 | Fact & Data Density | 6% | Specific numbers, statistics, data points per page |
+| Duplicate Content Blocks | 5% | Identical text blocks repeated across sections within a page |
 | Citation-Ready Writing | 4% | Self-contained definition sentences, single-claim statements |
+| Cross-Page Duplicate Content | 3% | Same paragraphs copy-pasted across multiple pages |
 | Answer-First Placement | 3% | Answer block in first 300 words, no throat-clearing openers |
 | Evidence Packaging | 3% | Inline citations, attribution phrases, sources sections |
 
@@ -64,9 +66,9 @@ AEORank evaluates 34 criteria that determine how AI engines (ChatGPT, Claude, Pe
 | Criterion | Weight | What it measures |
 |-----------|--------|------------------|
 | Direct Answer Paragraphs | 5% | Concise answer paragraphs after question headings |
-| Q&A Content Format | 5% | Question-format headings (What, How, Why) with answers |
-| Query-Answer Alignment | 5% | Every question heading followed by a direct answer |
-| Comprehensive FAQ Section | 4% | Dedicated FAQ with FAQPage schema markup |
+| Q&A Content Format | 4% | Question-format headings (What, How, Why) with answers |
+| Query-Answer Alignment | 4% | Every question heading followed by a direct answer |
+| Comprehensive FAQ Section | 3% | Dedicated FAQ with FAQPage schema markup |
 | Table & List Extractability | 3% | HTML tables with headers, ordered/unordered lists |
 | Definition Patterns | 2% | Clear "X is defined as..." patterns for key terms |
 | Entity Disambiguation | 2% | Primary entity defined early, consistent terminology |
@@ -107,19 +109,21 @@ AEORank evaluates 34 criteria that determine how AI engines (ChatGPT, Claude, Pe
 | RSS/Atom Feed | 1% | RSS feed linked from homepage |
 
 > **Coherence Gate:** Sites with topic coherence below 6/10 are score-capped regardless of technical perfection. A scattered site with perfect robots.txt, llms.txt, and schema will score lower than a focused site with mediocre technical implementation.
+>
+> **Duplication Gate:** Per-page scores are capped when duplicate content blocks are detected. A page with 3+ identical copy-pasted paragraphs cannot score above 35/75 regardless of other signals — LLMs will flag it as low-quality content.
 
 <details>
-<summary>All 34 criteria (numbered list)</summary>
+<summary>All 36 criteria (numbered list)</summary>
 
 | # | Criterion | Weight | Pillar |
 |---|-----------|--------|--------|
 | 1 | llms.txt File | 2% | AI Discovery |
 | 2 | Schema.org Structured Data | 3% | Trust & Authority |
-| 3 | Q&A Content Format | 5% | Content Structure |
+| 3 | Q&A Content Format | 4% | Content Structure |
 | 4 | Clean, Crawlable HTML | 2% | Technical Foundation |
 | 5 | Entity Authority & NAP Consistency | 5% | Trust & Authority |
 | 6 | robots.txt for AI Crawlers | 2% | AI Discovery |
-| 7 | Comprehensive FAQ Section | 4% | Content Structure |
+| 7 | Comprehensive FAQ Section | 3% | Content Structure |
 | 8 | Original Data & Expert Analysis | 10% | Answer Readiness |
 | 9 | Internal Linking Structure | 4% | Trust & Authority |
 | 10 | Semantic HTML5 & Accessibility | 2% | Technical Foundation |
@@ -136,7 +140,7 @@ AEORank evaluates 34 criteria that determine how AI engines (ChatGPT, Claude, Pe
 | 21 | Content Publishing Velocity | 2% | AI Discovery |
 | 22 | Schema Coverage & Depth | 1% | Technical Foundation |
 | 23 | Speakable Schema | 1% | Technical Foundation |
-| 24 | Query-Answer Alignment | 5% | Content Structure |
+| 24 | Query-Answer Alignment | 4% | Content Structure |
 | 25 | Content Cannibalization | 2% | AI Discovery |
 | 26 | Visible Date Signal | 2% | Technical Foundation |
 | 27 | Topic Coherence | 14% | Answer Readiness |
@@ -147,6 +151,8 @@ AEORank evaluates 34 criteria that determine how AI engines (ChatGPT, Claude, Pe
 | 32 | Entity Disambiguation | 2% | Content Structure |
 | 33 | Extraction Friction | 2% | Technical Foundation |
 | 34 | Image Context for AI | 1% | Technical Foundation |
+| 35 | Duplicate Content Blocks | 5% | Answer Readiness |
+| 36 | Cross-Page Duplicate Content | 3% | Answer Readiness |
 
 </details>
 
@@ -197,7 +203,7 @@ Or use `npx` directly:
 Run a complete audit. Returns `AuditResult` with:
 
 - `overallScore` - 0-100 weighted score
-- `scorecard` - 28 `ScoreCardItem` entries (criterion, score 0-10, status, key findings)
+- `scorecard` - 36 `ScoreCardItem` entries (criterion, score 0-10, status, key findings)
 - `detailedFindings` - Per-criterion findings with severity
 - `opportunities` - Prioritized improvements with effort/impact
 - `pitchNumbers` - Key metrics (schema types, AI crawler access, etc.)
@@ -219,10 +225,10 @@ Run a complete audit. Returns `AuditResult` with:
 
 ### `scorePage(html, url?)`
 
-Score a single HTML page against 14 per-page AEO criteria. Returns `PageScoreResult` with:
+Score a single HTML page against 21 per-page AEO criteria. Returns `PageScoreResult` with:
 
-- `aeoScore` - 0-100 weighted score
-- `criterionScores` - 14 `PageCriterionScore` entries (criterion, score 0-10, weight)
+- `aeoScore` - 0-75 weighted score (capped; duplication gate may lower further)
+- `criterionScores` - 21 `PageCriterionScore` entries (criterion, score 0-10, weight)
 
 ### `scoreAllPages(siteData)`
 
@@ -382,21 +388,22 @@ console.log(crawlResult.discoveredUrls.length); // Total URLs found
 
 ## Per-Page Scoring
 
-AEORank scores each individual page (0-75) against the 20 criteria that apply at page level. Instead of only seeing "your site scores 62," you get "your /about page scores 45, your /blog/guide scores 72."
+AEORank scores each individual page (0-75) against the 21 criteria that apply at page level. Instead of only seeing "your site scores 62," you get "your /about page scores 45, your /blog/guide scores 72."
 
-The 20 per-page criteria follow the same pillar-first weighting as the site-level score:
+The 21 per-page criteria follow the same pillar-first weighting as the site-level score:
 
 | Pillar | Per-Page Criteria | Weight |
 |--------|-------------------|--------|
 | **Answer Readiness** | Original Data & Expert Content | 10% |
 | | Fact & Data Density | 6% |
+| | Duplicate Content Blocks | 5% |
 | | Citation-Ready Writing | 4% |
 | | Answer-First Placement | 3% |
 | | Evidence Packaging | 3% |
 | **Content Structure** | Direct Answer Paragraphs | 5% |
-| | Q&A Content Format | 5% |
-| | Query-Answer Alignment | 5% |
-| | FAQ Section Content | 4% |
+| | Q&A Content Format | 4% |
+| | Query-Answer Alignment | 4% |
+| | FAQ Section Content | 3% |
 | | Table & List Extractability | 3% |
 | | Definition Patterns | 2% |
 | | Entity Disambiguation | 2% |
@@ -409,9 +416,11 @@ The 20 per-page criteria follow the same pillar-first weighting as the site-leve
 | | Image Context for AI | 1% |
 | **AI Discovery** | Canonical URL Strategy | 1% |
 
-The remaining 14 criteria are site-level only: llms.txt, robots.txt, sitemap, RSS, entity consistency, internal linking, content licensing, author schema, content velocity, schema coverage, speakable schema, content cannibalization, topic coherence, and content depth.
+The remaining 15 criteria are site-level only: llms.txt, robots.txt, sitemap, RSS, entity consistency, internal linking, content licensing, author schema, content velocity, schema coverage, speakable schema, content cannibalization, cross-page duplication, topic coherence, and content depth.
 
 > **Single-page cap:** Per-page scores are capped at 75 since single pages cannot demonstrate site-wide signals like topic coherence, content velocity, or sitemap completeness.
+>
+> **Duplication gate:** Pages with significant duplicate content blocks are score-capped. A page with 3+ copy-pasted paragraphs is capped at 35/75 — LLMs treat repeated content as low-quality regardless of other signals.
 
 ### CLI Output
 
@@ -436,7 +445,7 @@ import type { PageScoreResult, PageCriterionScore } from 'aeorank';
 // Score a single page
 const result = scorePage(html, url);
 console.log(result.aeoScore);         // 0-75 (capped for single pages)
-console.log(result.criterionScores);  // 20 per-criterion scores
+console.log(result.criterionScores);  // 21 per-criterion scores
 console.log(result.scoreCapped);      // true if score was capped at 75
 
 // Score all pages from site data
@@ -565,9 +574,13 @@ console.log(result.comparison.tied);              // Criteria with equal scores
 
 ## Changelog
 
+### v3.1.0 - Duplicate Content Detection
+
+2 new criteria (#35-#36): Duplicate Content Blocks (intra-page, 5%) and Cross-Page Duplicate Content (3%). Detects identical text blocks within pages and copy-pasted paragraphs across pages using shingle-based Jaccard similarity. Boilerplate filtering excludes CTAs, signups, and template content from false positives. Duplication gate caps per-page scores when severe duplication is found. CLI now shows duplicate section names inline per page.
+
 ### v3.0.0 - 5-Pillar Framework & 6 New Criteria
 
-Scoring Engine v2: 28 → 34 criteria with 5-pillar framework (Answer Readiness, Content Structure, Trust & Authority, Technical Foundation, AI Discovery). 6 new criteria targeting citation quality, evidence packaging, and extraction friction. Per-pillar sub-scores, top-3 fixes, client-friendly names. Single-page score cap at 75. 15 per-page quality checks (up from 12).
+Scoring Engine v2: 28 → 34 criteria (now 36) with 5-pillar framework (Answer Readiness, Content Structure, Trust & Authority, Technical Foundation, AI Discovery). 6 new criteria targeting citation quality, evidence packaging, and extraction friction. Per-pillar sub-scores, top-3 fixes, client-friendly names. Single-page score cap at 75. 15 per-page quality checks (up from 12).
 
 ### v2.3.0 - Coherence Scaling & Script Stripping
 
@@ -595,7 +608,7 @@ Individual page scores (0-100) against 14 page-level criteria. Top/bottom page r
 
 ## Benchmark Dataset
 
-The `data/` directory contains the largest open dataset of AI visibility scores - **13,619 domains** scored across 34 criteria, including **4,328 Y Combinator startups** across 48 batches (W06-W26):
+The `data/` directory contains the largest open dataset of AI visibility scores - **13,619 domains** scored across 36 criteria, including **4,328 Y Combinator startups** across 48 batches (W06-W26):
 
 | File | Contents |
 |------|----------|

package/dist/browser.d.ts

@@ -64,7 +64,7 @@ declare function buildLinkGraph(pages: FetchResult[], domain: string, homepageUr
 
 /**
  * V2 Pillar Framework — 5-pillar scoring model.
- * Maps all 34 criteria into pillars, computes sub-scores,
+ * Maps all 36 criteria into pillars, computes sub-scores,
  * provides client-friendly names, and calculates top-3 fixes.
  */
 
@@ -320,7 +320,7 @@ interface SitemapDateAnalysis {
 declare function countRecentSitemapDates(sitemapText: string): SitemapDateAnalysis;
 declare function extractRawDataSummary(data: SiteData): RawDataSummary;
 /**
- * Run all 34 criteria checks using pre-fetched site data.
+ * Run all 36 criteria checks using pre-fetched site data.
  * All functions are synchronous (no HTTP calls) - data was already fetched.
  */
 declare function auditSiteFromData(data: SiteData): CriterionResult[];
@@ -456,7 +456,7 @@ declare function analyzeAllPages(siteData: SiteData): PageReview[];
 
 /**
  * Per-page AEO scoring.
- * Evaluates 20 of 34 criteria that apply at individual page level.
+ * Evaluates 21 of 36 criteria that apply at individual page level.
  * Produces a 0-75 AEO score per page (single-page cap at 75).
  */