Core principle: write for retrieval
AI engines in search mode (ChatGPT Search, Perplexity, AI Overviews) work in two steps: a retrieval stage that pulls relevant passages from a corpus, then a generation stage that synthesises a response while citing those passages. Optimising for retrieval means making each of your paragraphs readable out of context.
Chunking: the granularity that matters
Chunk size, overlap and boundaries depend on the retrieval implementation. The original RAG research does not define a universal HTML segmentation rule for commercial answer engines.
| HTML component | Role in chunking | Best practice |
|---|---|---|
| H2 | Main section | One H2 = one distinct intent, with its implicit long-tail query. |
| H3 | Subsection | Sub-question or sub-aspect, never decorative. |
| Paragraph | Reading unit | One coherent idea, with the context and evidence it needs. |
| List | Near-extractable as-is | Standalone items, no "see above" references. |
| Table | Structured comparison | Clear headers, short cells, avoid merged cells. |
Standalone passages: test each one
Simple test: copy any paragraph of your page and paste it into an empty message to a colleague. If the paragraph stays understandable, it's standalone.
- Avoid pronouns without an antecedent ("it enables..." mid-page).
- Re-name the main entities at the start of each section.
- Define acronyms at their first local occurrence, not only at page top.
- Date time-bound statements ("in 2026", not "this year").
Citation-friendly content
A useful passage makes its claim verifiable. These editorial practices do not guarantee that an engine will select or cite it:
- A sharp claim, a dated statement linked to the relevant product documentation can be checked. "AI is changing SEO" isn't.
- Minimum context, who, what, when. No ambiguity on the subject.
- Verifiability, an external source, a published datum, an author.
Entities and disambiguation
LLMs bind your content to entities. If your brand shares its name with something else (a plant, a person, another company), disambiguation is priority one. Techniques:
- Systematic co-occurrence with domain markers: sector, product, customer segment.
- Foundational links to verified profiles representing the same entity, via
sameAsonOrganizationschema. - Factual biography on an About page with dates, places, activities, sources.
- Editorial consistency: same tone, same terminology across site and adjacent channels (LinkedIn, press, podcasts).
Anatomy of a GEO page
This is an editorial template, not a search-engine requirement. Adapt the sections and formats to the reader’s question.
- H1, a precise description of the main subject.
- Lede, a direct answer with enough context. First sentence standalone.
- Dates, publication + last update, visible.
- H2 "In brief", a summary only where it improves reading.
- Body, sections covering useful sub-questions.
- Table or checklist, use when it explains a comparison or sequence.
- Contextual FAQ, real questions that need a distinct answer.
- Outbound linking, links to the explanations and evidence the reader needs.
- Author and organisation, schema.org
Article+Organization.
Length, format, density
There's no magic length. A page must cover its subject, not hit a word quota. Benchmarks:
- Pillar: cover the main decision and direct readers to useful subtopics.
- Satellite: answer a distinct intent without repeating the main guide.
- FAQ / definition: answer fully without a word quota.
Common mistakes observed
- Walls of text, paragraphs that obscure the reasoning or context.
- Decorative H2s: "Conclusion", "Introduction", "Learn more", can be made more descriptive when appropriate.
- JSON-LD schemas inconsistent with visible content (missing author, fake date, wrong type).
- Unreviewed AI-generated content, stacking empty phrasing.
- Cross-page duplications that dilute authority.
- Long conditional sentences saying nothing citable.
Express checklist
- Each H2 carries a clear intent and reformulates a query.
- Each paragraph can be read in isolation.
- Every numerical claim is dated and sourced.
- Every acronym is defined at first occurrence.
- The page contains at least one table or checklist.
- The page carries a visible update date.
- Internal links lead to useful supporting explanations.
- schema.org structured data is validated.
Primary sources and scope
This edition applies the reviewed French guidance on content structure. Sources checked on 8 September 2026:
- Google: AI features and your website. SEO fundamentals remain relevant; there is no special AI markup requirement.
- Google: helpful, reliable content. Content should serve readers, without a preferred word count.
- Lewis et al.: Retrieval-Augmented Generation. A research architecture, not a universal specification for every answer engine.