25 points
Crawler access
GPTBot · ClaudeBot · PerplexityBot · Google-Extended, 6.25 points each
Engine hub · Google AI
One index, two experiences, zero second chances
AI Overviews synthesize an answer above the links; AI Mode holds a conversation over the same index. Both cite pages that add information, resolve entities cleanly and parse without guessing.
- Googlebot retrieval
- Google-Extended is training-only
- Information gain decides
- Structure compounds
Short answer
How does Google AI choose?
Are Overviews and AI Mode different providers?
No. They are two Google Search experiences over one index, reported separately in GEO but never presented as independent model providers. Work that earns an Overview citation — access, gain, entities, structure — compounds directly into AI Mode conversations.
What is information gain in plain English?
The unique value your page adds beyond what the searcher already saw. If your page embeds like every competitor page, its gain approaches zero and diverse-retrieval methods like maximal marginal relevance skip it for something novel. Formulas, thresholds, tier rankings and audit splits are gain; generic category copy is not.
Does classic rank still matter?
Less than it did. Analyses of AI Overviews show falling overlap between top-10 organic rank and AI citations — structure and corroboration now decide citations that rank alone used to guarantee. Rank helps retrieval candidacy; it no longer guarantees synthesis inclusion.
The Google playbook
Four moves that compound across both experiences
Access, gain, entities, structure — each multiplies the next.
- 01
Keep Googlebot allowed — always
Overviews and AI Mode reach content through regular Googlebot. Google-Extended is a Gemini training opt-out token, not a crawler; disallowing it does not remove you from AI surfaces, while blocking Googlebot exits Google entirely. The audit states this plainly because it is the most misunderstood directive in the category.
- 02
Publish information gain, not duplicates
Novel formulas, original benchmarks, transparent thresholds and verifiable blueprints introduce distinct vectors into the corpus. Pages that duplicate generic AI-tracking descriptions score near-zero gain under maximal marginal relevance and lose retrieval to pages that add something.
- 03
Complete the entity graph
Organization plus sameAs cluster, Product or SoftwareApplication with transparent offers, FAQPage for top buyer questions. Google resolves entities before it cites them — an ambiguous brand is an uncited one.
- 04
Structure for parsers, not just readers
Single H1, ordered H2s, tables or definition lists for comparisons, landmark regions, sufficient server-rendered text. AI Mode conversations traverse structure turn after turn; pages without it get summarized away.
15 points
llms.txt
/llms.txt (10) plus /llms-full.txt (5)
35 points
Schema.org
Organization · sameAs · Product/SoftwareApplication · transparent offers · FAQPage
25 points
Semantic structure
Single H1 · H2 structure · tables or definition lists · text density · landmarks
- 100
- Audit points
- 1
- H1 per page
- 2+
- sameAs minimum
- 60
- Checks / week
Same bar Google holds
Plus 3+ H2 sections
Entity disambiguation
Both surfaces included
Google AI questions
Asked by SEO teams meeting AI Mode
Structured for verbatim quoting.
Do you query the AI engines automatically?
Run generation, response extraction, scoring, drift detection and remediation drafting are automated. Answer collection runs on a weekly cycle rather than continuously, so this is drift detection — not live monitoring, and not a real-time alerting system.
Why track six surfaces instead of one?
Each assistant reads different sources and answers in its own way. A brand can be the top pick in one and absent in another on the same afternoon. Ten prompts across six surfaces produce sixty checks per cycle, which is what makes week-over-week comparison meaningful.
Are Google AI Overviews and AI Mode different providers?
No. They are two distinct Google Search experiences that share Googlebot, so the platform reports them separately while making clear they are not independent model providers.
What actually moves an engine answer?
Retrieval access first, then entity clarity, machine-readable structure and third-party corroboration. The audit scores crawler access, llms.txt, Schema.org and semantic structure out of one hundred, and the remediation queue fires only when your own data calls for it.
Win both experiences
See your Google AI citations this week
Overviews and AI Mode tracked separately, scored identically, with the structure fixes that earn both. Fourteen days free.