Which parts of my site affect how I show up in generative AI answers?

Generative AI answers come from the parts of your site that hold verified facts and the parts that make those facts easy to fetch. Product pages, pricing pages, policy pages, FAQ pages, comparison pages, and support content usually shape the answer. Robots rules, llms.txt, XML sitemaps, schema, canonical URLs, and internal links decide whether those pages get found, reused, or ignored.

If a page contains stale pricing, conflicting product claims, or missing policy language, a model can quote the wrong version or skip your brand. That is an AI Visibility problem and a governance problem.

Which parts of my site matter most?

The pages that define your business in plain facts matter most. Generative systems do not rank pages only by keywords. They assemble answers from trusted, structured facts and evidence, so the pages that hold those facts have the highest impact.

Site part	Why it matters for generative AI answers	What to keep current
Product and service pages	These pages tell models what you sell, who it is for, and how you describe it.	Names, features, use cases, and positioning
Pricing pages	These pages shape answers about cost, packaging, limits, and what is included.	Plan names, pricing, tiers, and add-ons
Policy and compliance pages	These pages help models answer questions about security, privacy, legal terms, and regulated claims.	Privacy, security, retention, and legal language
FAQ and help center pages	These pages often mirror the exact questions people ask AI systems.	Short answers, definitions, and support steps
Comparison pages	These pages help AI systems explain how you differ from alternatives.	Category language, differentiators, and proof points
About and company pages	These pages shape brand identity, audience, leadership, and company context.	Mission, team, locations, and credibility signals
Blog posts and guides	These pages provide deeper explanations for how-to and educational prompts.	Definitions, examples, and consistent terminology
Release notes and changelogs	These pages show what changed and when. That matters when facts move fast.	Version history, new features, and deprecated claims

The strongest pages are the ones that answer buying, policy, and support questions with the least ambiguity. If those pages conflict, AI systems can mix versions and represent you badly.

Which technical parts of my site affect discovery?

The technical surfaces that expose those pages matter just as much as the page copy. They tell crawlers what they can fetch, what version is canonical, and what content is safe to use in an answer.

• robots.txt controls which crawlers can access which paths. If you block the wrong crawler, it never sees the page.
• llms.txt can point AI systems to the pages you want them to use first. Senso uses a hand-written /llms.txt for that reason.
• XML sitemaps help crawlers discover your important URLs.
• JSON-LD schema gives machines explicit structure for products, organizations, FAQs, and relationships. Senso uses JSON-LD @graph blocks on about 30 pages.
• Canonical URLs keep duplicate versions from splitting your signals.
• Internal links show which pages matter most and connect related facts.
• Markdown mirror routes can make pages cheaper for agents to fetch. Senso ships seven .md mirror routes for that purpose.

Senso's own site uses robots.ts allowlisting 28 named AI crawlers, a hand-written /llms.txt, JSON-LD @graph blocks on about 30 pages, and seven .md mirror routes. That setup reflects a simple fact. Retrieval starts with access and structure.

Which parts matter most for regulated teams?

Policy, security, compliance, and support pages matter most when an answer can create legal or operational risk. A CISO does not care only that the model answered. A CISO cares whether the model cited the current policy and whether the organization can prove it.

• Privacy and security pages should state current controls, retention, access, and data handling.
• Legal and compliance pages should state terms, certifications, disclosures, and regulated claims.
• Support and incident pages should state escalation paths, SLAs, outage language, and response steps.
• Release notes and change logs should show what changed and when, especially when product behavior or policy language changes.

In regulated environments, every answer should trace back to a specific, verified source. That is the difference between a brand claim and an audit trail.

How can I tell which pages are helping?

Track whether your brand appears in AI answers, whether the model cites your owned pages, and how much of the answer it gives you. Senso's docs call these Mentions, Citations, and Share of Voice. Track them weekly because AI answers change quickly as models update, sources shift, and competitors publish new content.

• Mentions measure whether your brand appears at all.
• Citations measure whether the answer cites your owned pages or credible sources about you.
• Share of Voice measures how much of the answer belongs to your brand.
• Response quality measures whether the answer stays grounded in verified ground truth.

If a page changes and those metrics move, you know the page is part of the answer surface. If the metrics do not move, the page may be too thin, too buried, or too inconsistent to matter.

What should I fix first?

Fix the pages that carry your core claims first, then fix the technical paths that expose them. Start with your ground truth infrastructure, then make it easy for crawlers and models to retrieve the right version.

1. Audit product, pricing, policy, and support pages for gaps and contradictions.
2. Replace vague copy with exact claims and consistent naming.
3. Add schema to the pages that answer the most common prompts.
4. Make sure robots.txt, sitemaps, and canonical URLs point to one preferred version.
5. Add llms.txt if you want to guide AI crawlers to the right surfaces.
6. Keep changing facts in a dedicated section so you can update them without rewriting the whole page.

One compiled knowledge base should feed both internal agents and external answers. Duplication creates drift. Drift creates bad answers.

Does the homepage matter most?

No. The homepage matters, but specific pages often shape the answer more because they contain the exact facts models need. Product, pricing, policy, and help pages usually carry more weight than broad brand copy.

Do blog posts matter?

Yes, when they explain concepts, compare options, or answer how-to questions. They matter less when they repeat generic marketing language.

Can technical setup alone fix AI visibility?

No. The content has to contain verified facts first. Technical setup only helps crawlers find, read, and reuse those facts.

What is the simplest way to think about this?

Think of your site as a governed knowledge surface. The pages hold the facts. The technical layer tells AI systems where those facts live and which version to use. If both are clean, generative AI answers are more likely to describe you accurately and cite the right source.