TL;DR. Nine rules govern whether AI engines cite a page. They are not ranked by importance. They are jointly necessary: a page that fails any one of them is functionally invisible even if it passes the other eight. Use the table below as a pre-publish checklist.
The 9 rules
| # | Rule | What it demands | Where to learn |
|---|---|---|---|
| 1 | BLUF | 40–80 word direct answer in the first paragraph | Lesson 5.3 |
| 2 | Question-format headers | H2s phrased as the reader's actual question | Lesson 5.7 |
| 3 | Semantic HTML hierarchy | One H1, H2s, H3s, no skipped levels | Lesson 5.7 + Module 4.5 |
| 4 | Justification attributes | Pros/cons, comparison tables, "best for X" claims (see Sentiment tab) | Lesson 5.4 |
| 5 | Evidence and data | Named stats, dates, analyst sources, quantified claims | Lesson 5.6 |
| 6 | E-E-A-T signals | Named authors with credentials, bylines, review dates | Lesson 5.8 |
| 7 | JSON-LD schema | Article, FAQPage, HowTo, Organization, Product | Module 4.4 |
| 8 | Chunkability | 50–120 word self-contained paragraphs | Module 4.6 |
| 9 | Freshness signals | "Last updated" line, recent stats, dateModified schema | Lesson 5.5 |
The two rules that live in Module 4
Rules 7 (JSON-LD) and 8 (chunkability) are technical implementation. Module 4 covers them as SEO gaps because the diagnostic is structural, not editorial. Rule 3 (semantic hierarchy) sits in both modules: the technical diagnosis is in Module 4.5; the writing craft of question-format headers is in lesson 5.7.
The pre-publish checklist
Before you hit publish on any page meant to answer a target prompt, run this:
- First paragraph is a 40–80 word direct answer to the target prompt (Rule 1)
- Every H2 is phrased as a question (Rule 2)
- One H1, H2s, H3s, no skipped levels (Rule 3)
- Page contains at least one pros/cons list or comparison table (Rule 4)
- Every major claim has a named stat or source (Rule 5)
- Author is named with credentials, page has a review date (Rule 6)
- JSON-LD schema is valid in Google Rich Results Test (Rule 7)
- Every paragraph is 50–120 words and stands alone (Rule 8)
- Visible "Last updated: [Month Year]" line on the page (Rule 9)
Any "no" is a rewrite.
Why all 9 are jointly necessary
A page can rank on Google with 4 of 9 rules passing. It cannot get cited on ChatGPT with 8 of 9. AI engines do not weight rules; they gate on them. Miss one and the chunk gets deprioritized.
The exception: brand pages (homepage, About, product) often lean hardest on rules 4, 5, 6, 7. Blog content leans hardest on 1, 2, 8, 9. But the 9-rule minimum still applies everywhere.
Do this now:
Copy the pre-publish checklist into your team's CMS template or Notion. Require it on every page meant to answer a prompt you track. No checklist, no publish.
Start improving your AI visibility today with Rankscale.
Get started