Check this page with an assistantOpens a chat asking it to summarise this article and name the evidence behind each claim.
Claude opens with the prompt on your clipboard: Anthropic does not support prefilled prompts on the web, and we would rather copy it than ship a button that drops it.
Why the Vocabulary Is Confusing on Purpose
Some of these terms describe mechanisms search engines document. Some describe scores a tool vendor invented. Some are marketing labels applied to work that already had a name. All three get used in the same sentence, and the ambiguity is frequently commercial rather than accidental.
The distinction that matters most when someone is selling you something: is this a thing Google does, or a thing a tool computes? Crawling, indexing, and canonicalisation are documented. Domain authority and topical authority scores are models built by vendors. Both can be useful; only one is what the search engine is actually doing.
A–C
- AEO / GEO
Labels for optimising toward answer engines and generative results. The underlying work overlaps heavily with ordinary search work, which is worth knowing when it appears as a separate line item. Read more.
- AI Overviews
Generated summaries shown above results, served from the same index as ordinary results rather than a separate AI index. Read more.
- AI Mode
Google's conversational search surface, which decomposes a question into multiple searches before synthesising an answer. Read more.
- Anchor text
The visible, clickable words of a link. Descriptive anchors help readers, crawlers, and assistive technology; the published ratio tables are invented. Read more.
- Canonical
A hint declaring which URL should represent a set of duplicates. Google treats it as one signal among several and can select a different URL. Read more.
- Cannibalisation
Two or more of your pages competing for the same query in a way that leaves both weaker. Frequently diagnosed, less frequently real. Read more.
- Click depth
How many links from the homepage it takes to reach a page. As depth rises, crawl frequency and internal link equity fall. Read more.
- Core update
A broad reassessment of how content is evaluated, run several times a year. Not a penalty, and there is nothing to appeal. Read more.
- Core Web Vitals
Google's named loading, interactivity, and stability metrics. Real, measurable, and consistently overweighted relative to relevance. Read more.
- Crawl budget
The practical limit on how much of your site gets requested. A genuine constraint on large sites and essentially irrelevant on small ones. Read more.
- Crawling
A search engine requesting your pages. Distinct from indexing, and a page can be crawled without ever being indexed. Read more.
D–I
- Disavow
A tool for telling Google to ignore specific inbound links. Narrow, rarely appropriate, and most files are written out of anxiety rather than evidence. Read more.
- Domain authority (DA) / Domain rating (DR)
Third-party scores from tool vendors estimating backlink strength. Google does not read them and has never used them. Read more.
- Duplicate content
Substantially identical content at multiple URLs. There is no penalty; the costs are URL selection, split signals, and wasted crawling. Read more.
- E-E-A-T
Experience, expertise, authoritativeness, trustworthiness. A concept from the rater guidelines describing what Google aims to reward, not a ranking factor. Read more.
- Entity
A specific thing rather than a string: this company, this person, this product. Search systems reason about entities and their relationships. Read more.
- Faceted navigation
Filters that generate URL combinations. On large catalogues the single largest source of crawl waste. Read more.
- Featured snippet
An extracted answer shown above results. Far less common than it was, and the formatting that earned it now feeds generated answers. Read more.
- Hreflang
Markup connecting equivalent pages in different languages or regions. It connects equivalents; it does not create them. Read more.
- Impressions
How often your page appeared, whether or not anyone clicked. In the generative AI report, impressions are all you get. Read more.
- Index bloat
Far more URLs indexed than real pages, usually from parameters, archives, or filters. A reliable red flag when evaluating a site. Read more.
- Indexing
A search engine deciding to store and understand a page. Google states plainly that not every page it crawls will be indexed. Read more.
- Internal linking
Links between your own pages. The clearest statement you make about which page matters for a topic, and the cheapest fix available. Read more.
J–R
- llms.txt
A proposed file listing key pages for language models. No major provider has committed to reading it; documentation sites are the one arguable case. Read more.
- Long tail
Specific, low-volume queries that collectively make up most of search. Frequently reported as zero volume because the tools cannot see them. Read more.
- Meta description
A candidate for the snippet under your result. Not a ranking factor, and frequently replaced by text Google takes from the page. Read more.
- Noindex
A directive removing a page from results while leaving it crawlable. Distinct from a robots.txt block, which prevents crawling entirely. Read more.
- Nofollow / sponsored / ugc
Attributes describing why a link exists. Treated as hints rather than directives since 2020. Read more.
- Programmatic SEO
Generating pages from structured data at scale. Works where each page carries genuine data density, fails where it does not. Read more.
- Query fan-out
Google's term for AI Mode breaking a question into subtopics and issuing many searches at once. The sub-query counts circulating are inference. Read more.
- Rendering
Executing JavaScript to produce the final page. A separate step that can happen later than the initial crawl. Read more.
- Rich result
An enhanced search listing produced from structured data. The set of types that still earn one has been shrinking for three years. Read more.
- Robots.txt
A file requesting that crawlers avoid certain paths. A request honoured by well-behaved crawlers, not an enforcement mechanism. Read more.
S–Z
- Schema / structured data
Markup labelling what a page visibly contains. It describes content that exists; it does not assert content that does not. Read more.
- Search intent
What someone is trying to accomplish, read from the results page rather than guessed from the keyword. The most common reason sound pages fail. Read more.
- Semantic search
Retrieval based on meaning rather than string matching. Why a page can rank for words it never contains. Read more.
- Service area business
A business that travels to customers rather than receiving them. Requires the address to be hidden if there is no signed premises. Read more.
- Share of model
How often an assistant names you on category questions, relative to competitors. Measurable by repeated sampling, not by a vendor score. Read more.
- Site reputation abuse
Publishing content on a domain mainly to borrow its standing. Enforcement is algorithmic and editorial improvement is not a route back. Read more.
- Sitemap
A file listing your URLs for discovery. Generating one and submitting it are two different things, and the gap is a common failure. Read more.
- Topical authority
Shorthand for covering a subject completely and connecting it clearly. Not a published score, whatever tools report. Read more.
- YMYL
Your Money or Your Life: topics where inaccuracy causes real harm. A description of stakes from the rater guidelines, not a penalty class. Read more.
- Zero-click search
A search resolved without anyone visiting a site. Changes what winning a query is worth rather than how you win it. Read more.
Terms Worth Treating Carefully
A few of these come up most often in sales conversations, and knowing what they are not is more useful than knowing what they are.
| Term | Commonly implied | Actually |
|---|---|---|
| Domain authority | A Google score for your site | A vendor model Google does not read |
| E-E-A-T | A ranking factor to optimise | A rater-guideline concept describing what is being aimed at |
| Duplicate content penalty | A sanction for repeated text | Does not exist; the cost is URL selection |
| Topical authority score | Something Google assigns | Shorthand for coverage; the score is a tool's |
| llms.txt optimisation | A visibility service | A file with no confirmed major consumer |
| Google penalty | Any traffic decline | A manual action, which appears in Search Console by name |
The last row causes the most wasted effort. Most declines are core update reassessments, seasonality, or something that shipped, none of which is a penalty and none of which has an appeal. The diagnostic order is in page-level authority and SEO beliefs that were true once.
Questions About SEO Terminology
- Which SEO terms describe things Google actually does?
Crawling, indexing, canonicalisation, rendering, structured data, and core updates are all documented mechanisms with published guidance. Domain authority, topical authority scores, and most acronyms ending in EO are either vendor metrics or industry shorthand, which does not make them useless but does mean Google is not reading them.
- Why do so many SEO terms mean different things?
Because the field mixes three vocabularies: documented mechanisms from search engines, metrics invented by tool vendors, and marketing labels invented by agencies. All three get used in the same sentence, and the confusion is frequently commercial rather than accidental.
- What is the difference between crawling and indexing?
Crawling is a search engine requesting your page. Indexing is it deciding to store and understand that page so it can be shown. The two are separate and a page can be crawled and never indexed, which is one of the most common and most misdiagnosed states a page can be in.
- Are AEO, GEO, and LLMO real disciplines?
AEO, GEO, and LLMO are labels for optimising toward AI answer surfaces, and the underlying work overlaps heavily with ordinary search work. The distinction is worth knowing because the acronyms are frequently used to justify new line items for things that were already being done under a different name.
- How current is this glossary?
It reflects mid-2026, including the structured data types Google has retired, the AI surfaces that now exist, and the metrics that have been formally deprecated. Terminology in this field ages quickly, and definitions that were accurate in 2022 are frequently wrong now rather than merely dated.
