Knowledge hub
Practical, source-checked guides for the parts of modern visibility that can actually be observed: crawl access, indexability, structured data, entity clarity, topical coverage, AI mentions and cited URLs. The hub separates documented search guidance from emerging GEO practice, and avoids invented AI ranking claims.
The words matter
Each step below is a separate event with a separate cause. None of them makes the next one happen. Most bad advice in this field comes from collapsing two of them together.
Updated baseline
Google documents no additional requirements to appear in AI Overviews or AI Mode, and no special markup or AI text file to add. That is Google speaking about its own AI features, not a general law about answer engines, but it does mean the work starts with ordinary search fundamentals.
Being crawlable, indexable and clearly written is what makes a page eligible to be retrieved and quoted. None of it obliges any system to mention you. No provider documents how grounding selection actually works, and anyone who tells you otherwise is guessing.
Operators now run separate agents for search indexing, for training corpora, for ad safety checks and for fetches a person triggers in a chat window. Each needs its own robots.txt rule, and a user-triggered agent may not consult robots.txt at all, so one blanket line rarely says what you mean.
Mention rate, citation rate and share of voice mean something only when the prompt set, provider, model version, date, denominator and cited URLs are all recorded. A single AI answer is not a ranking report.
Seven guides that set up the vocabulary everything else assumes. Read them in order if the difference between crawled, retrieved, cited and recommended is not yet second nature.
GEO is the practice of making a site retrievable and citable by AI answer engines. Here is what it is, what it is not, and what the evidence supports.
Most of what works in AI search is ordinary SEO. This article shows where the two overlap, where retrieval and citation add new steps, and what is still unproven.
Thirty checks, ordered by how much evidence stands behind them — from published standards down to ideas that are not fixes and must not be scored.
What an AI visibility tool can honestly observe, how controlled prompts and provider APIs produce it, and why a "ChatGPT rank" is not a measurement.
Crawled, retrieved, used, cited, recommended — five distinct steps, each with its own preconditions and its own evidence. Conflating them is costly.
Not every AI bot does the same job. Here are the documented categories, the agents each company publishes, and what blocking one actually costs you.
Schema.org is a real standard with real search uses and no documented role in AI citation. Here is where it earns its place and where the claims run out.
The groundwork that decides whether a page can be discovered, crawled, indexed and understood at all. Google's own guidance for its AI features points back to this foundation rather than to a separate discipline, so this is where the work starts.
A step-by-step audit of the things that decide whether machines can reach, parse and understand a WordPress site — with the evidence tier for each check.
Four tool categories, four different jobs: what on-page SEO plugins, rank trackers, site auditors and AI visibility monitors each actually produce.
What Google actually says internal links do, how anchor text carries meaning, how to find orphan pages, and which retrieval claims are evidence-backed and which are not.
Freshness is a property of some topics, not all. When an update earns its cost, when it does not, and how to keep dates honest.
What a pillar page is and is not, how a topic cluster is built and linked, and how SiteRank AI scores pillar candidates from links, relevance and length.
Generative engine optimisation is about making a page easy to retrieve, quote and attribute. It can improve eligibility. It cannot make a model mention or cite you, and nothing in this section is a confirmed ranking factor.
GEO is the practice of making a site retrievable and citable by AI answer engines. Here is what it is, what it is not, and what the evidence supports.
Most of what works in AI search is ordinary SEO. This article shows where the two overlap, where retrieval and citation add new steps, and what is still unproven.
Thirty checks, ordered by how much evidence stands behind them — from published standards down to ideas that are not fixes and must not be scored.
Google runs a passage ranking system and retrieval systems index chunks. Here is how to structure content for that without writing artificially for machines.
One name, one URL, one identity, expressed consistently in prose and in markup. What is standard, what is emerging, and what is guesswork.
No provider documents how a model chooses what to cite. This article separates the factors with real evidence behind them from the ones that are plausible guesswork.
Google says there are no additional requirements or special optimisations for its AI features. Here is what that means, and which controls actually exist.
Which bot is which, what each one is for, and how to write a policy that says what you actually mean. Operators run separate agents for search indexing, for training corpora and for fetches a person triggers in a chat window, and Perplexity documents that its user-triggered agent generally does not follow robots.txt at all. RFC 9309 is explicit that these rules are not access authorisation and no substitute for content security, and that listing a path in robots.txt advertises it to anyone who reads the file.
Not every AI bot does the same job. Here are the documented categories, the agents each company publishes, and what blocking one actually costs you.
What robots.txt can and cannot do for AI crawlers, worked policies you can copy, and how WordPress serves the file virtually.
OpenAI runs four documented agents with independent robots.txt settings. Here is what each does, and exactly what blocking it costs you.
Anthropic runs three bots and states why IP blocking is the wrong opt-out. Here is each one's purpose and the documented cost of disabling it.
Perplexity runs two agents, and states that one of them generally ignores robots.txt. Here is what each does and how to configure for it.
A licensing decision, not an SEO one. What training crawlers are, what blocking actually changes, and a framework for deciding by site type.
llms.txt is a community proposal for a Markdown site map aimed at LLMs. Here is the format, its real status, and what publishing one does not do.
Schema.org markup describes what is already on a page so a machine does not have to infer it. Google states that correct markup does not guarantee a rich result, and that no special structured data is needed to appear in its AI features. Entity clarity is the same problem in plain language: is it obvious who you are and what you do?
Schema.org is a real standard with real search uses and no documented role in AI citation. Here is where it earns its place and where the claims run out.
Organization markup, sameAs, and consistent identity details make a site's publisher unambiguous to machines. What that buys you, and what it does not.
One name, one URL, one identity, expressed consistently in prose and in markup. What is standard, what is emerging, and what is guesswork.
Google says there are no additional requirements or special optimisations for its AI features. Here is what that means, and which controls actually exist.
How to observe whether AI systems mention your site, under a prompt set, a provider, a model version and a date you have written down. Every figure here is an observation with a denominator attached. None of it is a rank, and a single answer is not a report.
What an AI visibility tool can honestly observe, how controlled prompts and provider APIs produce it, and why a "ChatGPT rank" is not a measurement.
Exact numerators, denominators and confidence intervals for mention rate and citation rate, with worked numbers and the limits of each figure.
Share of voice is a proportion inside a competitor set you define. Here is how the set changes the number, and how to choose one that stays honest.
Building a prompt library, sorting it into intent clusters, sampling enough to mean something, and keeping the cost of an AI visibility programme under control.
An API observation is not a consumer chat answer. Here are the seven layers that differ, and why we label the measurement instead of claiming a ranking.
A citation is the narrower and harder signal: a specific URL of yours shown as a source. These guides cover what makes a page worth citing, how citation sets drift between runs, and how to compare yourself against competitors without inventing a league table.
Crawled, retrieved, used, cited, recommended — five distinct steps, each with its own preconditions and its own evidence. Conflating them is costly.
No provider documents how a model chooses what to cite. This article separates the factors with real evidence behind them from the ones that are plausible guesswork.
Why the same prompt cites different sources next week, what actually moved, and how to tell a real change from sampling noise.
Share of voice in AI answers is a proportion within your own prompt set and your own competitor list. Here is the exact formula and what it cannot mean.
Coverage and connectivity: whether your site addresses a subject completely, and whether the pages that do so can actually be reached. Both are properties you can inspect directly on your own site, without asking anyone else's system what it thinks of you.
Topical authority is not a score any engine publishes. It is coverage, clustering, connectivity and depth — properties you can compute and act on.
What a pillar page is and is not, how a topic cluster is built and linked, and how SiteRank AI scores pillar candidates from links, relevance and length.
What Google actually says internal links do, how anchor text carries meaning, how to find orphan pages, and which retrieval claims are evidence-backed and which are not.
Google runs a passage ranking system and retrieval systems index chunks. Here is how to structure content for that without writing artificially for machines.
Freshness is a property of some topics, not all. When an update earns its cost, when it does not, and how to keep dates honest.
The whole library in publication order, for anyone who would rather browse than be guided.
Welcome to WordPress. This is your first post. Edit or delete it, then start writing!
Four tool categories, four different jobs: what on-page SEO plugins, rank trackers, site auditors and AI visibility monitors each actually produce.
llms.txt is a community proposal for a Markdown site map aimed at LLMs. Here is the format, its real status, and what publishing one does not do.
GEO is the practice of making a site retrievable and citable by AI answer engines. Here is what it is, what it is not, and what the evidence supports.
Topical authority is not a score any engine publishes. It is coverage, clustering, connectivity and depth — properties you can compute and act on.
Organization markup, sameAs, and consistent identity details make a site's publisher unambiguous to machines. What that buys you, and what it does not.
Schema.org is a real standard with real search uses and no documented role in AI citation. Here is where it earns its place and where the claims run out.
A licensing decision, not an SEO one. What training crawlers are, what blocking actually changes, and a framework for deciding by site type.
Share of voice in AI answers is a proportion within your own prompt set and your own competitor list. Here is the exact formula and what it cannot mean.
Most of what works in AI search is ordinary SEO. This article shows where the two overlap, where retrieval and citation add new steps, and what is still unproven.
What robots.txt can and cannot do for AI crawlers, worked policies you can copy, and how WordPress serves the file virtually.
Thirty checks, ordered by how much evidence stands behind them — from published standards down to ideas that are not fixes and must not be scored.