{
  "title": "Our blog archive checked against Google's generative AI search guidance",
  "run": "2026-10-15",
  "corpus": "Every post live on fifteenthmeridian.com/blog on the run date, excluding this post: body text and FAQ answers, split into sentences.",
  "source": "Our own blog archive",
  "rules": [
    {
      "id": "llms-txt",
      "name": "llms.txt mentioned",
      "pattern": "/llms\\.txt/i"
    },
    {
      "id": "chunking",
      "name": "Chunking content for machines",
      "pattern": "/\\bchunk(?:s|ing|ed)?\\b|break (?:your |the )?(?:content|page|post)s? into (?:small|tiny|short)/i"
    },
    {
      "id": "ai-writing",
      "name": "Writing specifically for AI",
      "pattern": "/\\bwrit(?:e|ing) (?:specifically |differently )?for (?:ai|llms?|chatgpt|answer engines|machines)\\b|ai-specific writing/i"
    },
    {
      "id": "schema-required",
      "name": "Special schema presented as required for AI",
      "pattern": "/(?:schema|structured data)[^.]{0,80}\\b(?:required|must have|need(?:s|ed)? to|is essential)\\b[^.]{0,60}\\b(?:ai|answer engine|chatgpt|ai overviews?|ai mode|llm)/i"
    },
    {
      "id": "mentions",
      "name": "Chasing mentions across the web",
      "pattern": "/\\b(?:build|earn|get|chase|seek|seeking|chasing|buy)\\b[^.]{0,30}\\b(?:brand )?mentions\\b/i"
    },
    {
      "id": "fan-out",
      "name": "A page per query variation",
      "pattern": "/\\bpage (?:for|per) (?:every|each) (?:query|question|variation|keyword|search)|\\bpages? per (?:practice area|query|keyword|city)\\b|separate (?:pages?|content) for every/i"
    },
    {
      "id": "internal-metrics",
      "name": "A tool claimed to read Google internals",
      "pattern": "/(?:tool|platform|software|dashboard)[^.]{0,80}\\b(?:google'?s internal|internal google|access to google'?s|approved by google|google-approved)/i"
    }
  ],
  "hedge": {
    "pattern": "/\\boptional\\b|not a (?:ranking factor|citation (?:guarantee|requirement)|requirement)|no guarantee|(?:does not|doesn'?t|cannot|can'?t|won'?t|will not) guarantee|guarantees? nothing|(?:ignores?|ignored)|(?:does not|doesn'?t) (?:use|read)|neither harm nor help|harmless|unproven|no evidence|(?:does|did) nothing|not required|\\bproposal\\b|\\bproposed\\b|not (?:needed|necessary)|no (?:effect|known effect)|unclear|optional discovery/i",
    "note": "A llms.txt sentence counts as hedged when this pattern matches in the same sentence. The test is strict: link text and headings that name the file count as unhedged, so every flagged post was read in full by a person."
  },
  "manualRead": "Each post with a flag was read in full. Verdicts: hedged (the post presents llms.txt as optional, unproven or ignored by Google), neutral-mention (names the file only in a link, a method note or a count), implies-it-helps (presents the file as improving citations or visibility).",
  "code": "tools/guidance-checker.mjs in the site repository",
  "summary": {
    "readOn": "2026-10-15",
    "postsScanned": 84,
    "postsTaggedAeo": 27,
    "postsMentioningLlmsTxt": 24,
    "hedged": 19,
    "neutralMentionOnly": 1,
    "impliesItHelps": 4,
    "postsFlaggedByOtherRules": 11,
    "otherRulesConfirmedAfterManualRead": 0,
    "offTopicMisfires": 2
  },
  "afterTheRun": "The four posts with the verdict implies-it-helps were corrected on 2026-10-15, the same day as the run; flags.csv records their wording as it read when the checker ran.",
  "slug": "google-ai-search-guidance-audit-2026",
  "data_collected": "2026-10-15",
  "posts": 84,
  "headline": "Of 24 Meridian15 posts that mention llms.txt, 19 call it optional or unproven, 1 only names it and 4 implied it helps; the 4 were corrected.",
  "sample_label": "84 live Meridian15 blog posts, every sentence of body text and FAQ answers",
  "collected_label": "October 2026",
  "key_findings": [
    {
      "value": "24",
      "claim": "Posts that mention llms.txt"
    },
    {
      "value": "19",
      "claim": "Call it optional, unproven or ignored by Google"
    },
    {
      "value": "4",
      "claim": "Implied it helps, all corrected on 2026-10-15"
    },
    {
      "value": "0",
      "claim": "Posts advising chunking, AI-specific writing, required schema or chasing inauthentic mentions, after a manual read"
    },
    {
      "value": "11",
      "claim": "Posts flagged by the other six rules: 11, none confirmed after a manual read; 2 of them off-topic misfires"
    }
  ],
  "citation": "Meridian15 (2026). Google Says llms.txt Does Nothing for Search. We Checked Our Own 84 Posts. https://fifteenthmeridian.com/blog/google-ai-search-guidance-audit-2026",
  "reuse": "Quote any figure with a link to the post. The rules, the flagged sentences and the per-post verdicts are in this folder."
}