1. The two crawlers
Googlebot
The classic Search crawler. Never block it unless you don't want to rank on Google Search.
Google-Extended
Controls Gemini training + AI Overviews grounding. Blocking it removes you from Overviews but leaves Search rankings intact.
2. Winning AI Overview citations
- State the canonical answer to the query in the first 120 words of the page. Overviews extract from the top.
- Wrap the answer in a semantically obvious container — a definition list, a numbered how-to, a comparison table.
- Ship the answer in JSON-LD too (FAQPage, HowTo, DefinedTerm). Overviews prefer sources with corroborating schema.
- Cite primary sources in the visible page — Overviews cite pages that themselves cite.
- Use canonical, unambiguous entity names in H2s. Overviews resolve entities against the Knowledge Graph.
3. Signals Gemini keys on
Gemini's grounding layer prefers pages with high entity coverage, resolvable schema, and low ambiguity between title and body. Pages that hedge — "it depends" answers, list of options with no recommendation — get cited less than pages that commit.
Grounded retrieval favours recency for QDF-heavy queries and evergreen authority for stable ones. WBP's Up to Date plugin picks the right refresh cadence per URL based on the query class.
4. How WBP configures for Gemini
WBP's default robots.txt allows Google-Extended by default (opt-in on the setup wizard). The AI Tuning Score checks every URL for Overview-eligibility: first-120-word answer, matching schema, entity resolvability. Failures are flagged on the dashboard.