Legible to the models people ask now.
Answer Engine Optimization and Generative Engine Optimization: making your site legible to the systems more people now ask instead of searching. Nobody can promise you a citation, and anyone who does is guessing. What can be done is the work that makes you findable, quotable, and worth quoting.
An AI visibility pass, built on what we control.
Content structure.
- Direct-answer paragraphs on key pages
- Question-first heading hierarchy
- Definition blocks for industry terms
- Quotable claim formatting
Question and answer content.
- FAQ sections on service and contact pages
- Questions taken from real inquiries and search data
- Answers that stand on their own out of context
- FAQPage markup where it matches the visible page
Entity definition.
- Organization schema on every page
- Clear "what we are" definition statements
- knowsAbout and sameAs properties
- Consistent NAP and entity data
llms.txt configuration.
- llms.txt file at the site root
- A maintained index of your best content
- Site overview and key pages mapped
- Reviewed when pages change, so it stays true
AI crawler permissions.
- robots.txt written per crawler, search and training kept separate
- OAI-SearchBot, ClaudeBot, PerplexityBot, GoogleOther
- CDN and firewall rules checked, not just robots.txt
- Access verified against real crawler hits in your logs
Local & generative signals.
- LocalBusiness schema where relevant
- Service area and territory markup
- Topical authority pages on key terms
- Citation-ready About content
Three things worth doing regardless.
How any given model picks its sources is not published, and it changes without notice. So this page does not claim to know. What follows are three things that are defensible on their own terms, measurable, and would still be worth doing if every AI answer engine shut down tomorrow.
Can the crawlers reach the page at all?
This is the one part of AI visibility that is genuinely binary. A page behind a blocking robots.txt rule, a CDN challenge, or a firewall rule nobody remembers writing cannot be read, quoted, or summarized by anything. Access is also the part most often broken by accident, and the easiest to prove either way.
- Search and training crawlers handled separately
- CDN, WAF, and hosting rules checked, not just robots.txt
- Confirmed against real crawler hits in your logs
Does the page answer the question directly?
A page that opens with a plain definition of what the business does is easier for any reader to extract an answer from than one that opens with "We are a passionate team of…". That is true of people, search engines, and language models alike, and it is worth doing on its own merits rather than because of any claim about how models choose sources.
- Direct-answer paragraph under every question heading
- Plain definitional statements, early on the page
- Structured data that matches what the page visibly says
Is there something only you can say?
Generic pages restate what a hundred other pages already say, and nothing is gained by being the hundred and first. Specific, original, checkable material is what gives any system a reason to reach for your page rather than someone else's: real numbers, real constraints, real decisions, with a date and a name against them.
- Original detail drawn from your actual work
- Named authors and review dates on substantial pages
- Claims kept current when the underlying facts move