Technical SEO terms

Classical SEO encompasses three areas: technical SEO (crawlability, performance, structured data), on-page SEO (titles, headings, content, internal links), and off-page SEO (backlinks, mentions, reputation). Google has been the dominant target for two decades.

The rise of AI assistants and answer engines has not replaced SEO — it has extended it. Strong classical SEO is still a prerequisite for AI visibility, because most AI systems read the same web that Google indexes. AgenticSEO adds a second optimization target on top: the LLMs and answer engines that summarize and cite that web.

In classical SEO, keyword research drives page-level decisions: which terms a page targets, what its title and headings say, and which related concepts it should cover. Tools like Google Search Console, Semrush, and Ahrefs estimate the search volume and difficulty of each candidate keyword.

In AgenticSEO, the unit shifts from keyword to prompt. People do not type 'best CRM for startups' into ChatGPT — they describe their situation in a full sentence. Keyword research still matters, but it is paired with prompt research: the representative questions a target buyer is likely to ask an AI assistant.

The most consequential tags for discoverability are <title>, <meta name="description">, <link rel="canonical">, the OpenGraph and Twitter Card meta tags, and JSON-LD structured-data <script> tags. AgenticSEO adds a fourth layer of AI-specific meta tags (ai:summary, ai:keywords, ai:intent, ai:audience) intended to be lifted directly by LLMs.

A tag audit is one of the fastest, lowest-risk wins in an AgenticSEO program: it changes nothing about the visible page but materially improves how search engines and AI assistants represent it.

Without canonical tags, query-string variants, tracking parameters, and HTTP/HTTPS or www/non-www duplicates fragment a page's ranking signals. Canonical tags tell search engines and LLMs which URL is the source of truth.

Every indexable page should declare its own canonical URL. Pages should not declare a canonical that points elsewhere unless they are genuinely duplicates.

Core OpenGraph tags are og:title, og:description, og:url, og:type, and og:image. Twitter Cards extend the same idea with twitter:card, twitter:title, etc.

LLMs do not yet uniformly use OpenGraph data, but answer engines that pull rich-link previews into their UI (Perplexity, Bing Copilot) increasingly do. Treat OpenGraph as table-stakes hygiene on every public page.

FAQPage is one of the highest-leverage schema types for AgenticSEO. AI assistants disproportionately quote content marked up as FAQ — the question/answer structure aligns directly with how users phrase prompts.

A good FAQPage block has 3-8 questions, each with a one-paragraph answer. Questions should be phrased the way real users ask them, not the way marketers would write them.

Schema.org defines hundreds of types (Organization, Person, Product, Article, FAQPage, Recipe, Event, DefinedTerm, and many more) and the properties each type accepts. Pages declare which Schema.org type they represent using JSON-LD, Microdata, or RDFa.

In the AI era, Schema.org has become more important, not less. LLMs and answer engines rely on it to construct knowledge-graph entries about brands and to decide which pages can be cited as authoritative on a topic.

JSON-LD is the structured-data format Google, Bing, and most LLMs prefer. It ships inside a <script type="application/ld+json"> tag in the page head and describes the page as a typed entity — Organization, Product, FAQPage, Article, BreadcrumbList, DefinedTerm, and so on.

For AgenticSEO, JSON-LD is foundational. AI assistants and answer engines use it to disambiguate what a page is about, who published it, and which entities (people, places, products) it references. A page with valid JSON-LD is dramatically more likely to be cited correctly than a page without.

Validate JSON-LD with Google's Rich Results Test or Schema.org's validator. Invalid JSON-LD is silently ignored — worse than having none at all.

sitemap.xml lives at the root of a website and follows the sitemaps.org schema. Each <url> entry can include a last-modified date, change frequency, and priority hint. The file is referenced from robots.txt with a Sitemap: directive and submitted to Google Search Console and Bing Webmaster Tools.

For AgenticSEO, the sitemap is the canonical list of pages that AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended — will discover and index. A complete, current sitemap paired with a well-formed llms.txt is the foundation for everything else.