glossary
Plain-language definitions for SEO and AI-search terminology.
A
- AEO (Answer Engine Optimisation)↗
- The practice of structuring content to be retrieved and cited by AI answer engines such as ChatGPT Search, Perplexity, and Google's AI Overviews. Often used interchangeably with GEO; the emphasis is on earning inclusion in synthesised answers rather than ranked positions.
- Agentic Search↗
- AI agents performing multi-step research across multiple sources to answer complex queries. Distinct from single-query retrieval; agents can break questions into sub-queries, browse multiple pages, and synthesise information autonomously.
- Agentic SEO↗
- Using AI agents to execute SEO workflows autonomously: keyword research, content auditing, technical fixes, and implementation, with minimal human input at each step. Distinct from agentic search (optimising content for retrieval by AI research agents).
- AI agent↗
- Software that pursues a goal autonomously across multi-step tasks, using tools, memory, and reasoning to plan and act without human approval at each step. Distinct from AI assistants (which respond to requests per session) and chatbots (which follow pre-defined rules).
- AI Overview↗
- A generative AI answer shown at the top of Google's search results, synthesised from multiple sources with citations. Often produces zero-click outcomes for informational queries.
- Algorithm update↗
- A change to how a search engine ranks results. Google deploys broad core updates several times per year; smaller updates are continuous.
- Alt text↗
- The text alternative for an image, used by screen readers and search engines. Should describe the image accurately and concisely.
- Anchor text↗
- The clickable text of a hyperlink. Used by search engines as a relevance signal for the linked page; over-optimised exact-match anchors can trigger penalties.
- AS (Authority Score)↗
- Semrush's proprietary score (0-100) estimating a domain's overall SEO authority, combining link, traffic, and quality signals.
- AI Traffic
- Visits to a site from AI answer engines such as ChatGPT Search, Perplexity, Copilot, and Claude. Growing traffic source as these systems complement or replace traditional search.
- AI Visibility↗
- How frequently a site is cited or retrieved by AI answer engines across multiple queries. A complement to search rankings; two sites can have similar Google visibility with very different AI citation rates.
- Anchor Text Profile↗
- The distribution of anchor text across all backlinks pointing to a page. Over-optimised profiles (mostly exact-match keywords) can trigger penalties; natural profiles are diverse and mix branded, generic, and partial-match anchors.
- Author Box
- A bio module displaying the author's name, credentials, and expertise at the top or bottom of an article. Strengthens E-E-A-T signals and is recommended by Google for YMYL content.
B
- Backlink↗
- A hyperlink from another website pointing to yours. One of the strongest ranking signals; quality matters more than quantity.
- Broken Link Building↗
- A link-building tactic that finds broken links on relevant sites and offers replacement content. Lower barrier than pitching from scratch since the site owner already has a reason to link.
- Black hat SEO↗
- Tactics that violate search engine guidelines: cloaking, sneaky redirects, automated content generation, link schemes. Carries penalty risk.
- Bounce rate
- The percentage of visits where the user leaves without further interaction. A weak SEO signal in isolation; context dependent.
- Brand mentions↗
- Online references to a brand name, product, or person, whether linked or unlinked. Unlinked mentions can signal brand authority to search engines; linked mentions also pass link equity. Also called brand signals.
- Breadcrumbs
- A navigation aid showing the user's location within a site hierarchy. Often marked up with BreadcrumbList schema for SERP display.
C
- Canonical tag↗
- An HTML element (`<link rel="canonical">`) declaring the preferred URL when multiple URLs serve substantially the same content.
- Circle to Search↗
- An Android feature that allows users to initiate a Google search from within any app by circling, highlighting, scribbling on, or tapping content on screen: text, images, or video frames. The search runs without leaving the current app and uses Google's standard ranking systems.
- Citation (local)↗
- An online mention of a business including its name, address, and phone number (NAP). Important for local pack ranking.
- CLS (Cumulative Layout Shift)↗
- A Core Web Vitals metric measuring unexpected layout shifts during page load. Good: under 0.1.
- Cluster (content cluster)↗
- A group of pages covering related sub-topics, linked back to a central pillar page. Builds topical authority.
- Core Web Vitals↗
- Google's set of three user experience metrics used as ranking signals: LCP (load speed), INP (interactivity), and CLS (visual stability). Introduced as a ranking factor in 2021.
- Crawl budget↗
- The number of URLs a search engine crawler will fetch on a site within a given period. A constraint primarily for very large sites.
- Crawlability↗
- The degree to which search engine bots can access and fetch the pages of a site.
- ccTLD (Country-Code Top-Level Domain)
- A domain extension assigned to a specific country: .co.uk for the UK, .de for Germany, .fr for France. A strong geotargeting signal, but limits flexibility if a business expands internationally.
- Content audit↗
- A systematic review of a site's existing pages, assessing each for performance, accuracy, and strategic value. Informs decisions on whether to keep, update, merge, or remove pages.
- Content gap analysis↗
- The process of identifying topics or keywords that competitors rank for which a site does not cover. Used to prioritise new content that can capture existing search demand.
- Content pruning↗
- The practice of removing or consolidating low-value, thin, or underperforming pages to improve a site's overall quality signals and concentrate crawl budget on content that earns traffic.
- Crawl error
- A failure recorded when a search engine bot attempts to fetch a URL and receives an error response: a 404, server error, or DNS failure. Reported in Google Search Console.
- CTR (Click-Through Rate)
- The percentage of impressions that result in a click. A primary measure of SERP performance.
- Client-Side Rendering (CSR)↗
- Rendering HTML in the user's browser after JavaScript executes. Can delay content visibility to crawlers unless the site prerendering or hydrates correctly. Contrasts with Server-Side Rendering.
- Commodity content
- Content on high-volume, low-differentiation topics where many sites publish similar information with minimal unique perspective or data. Difficult to rank because competition is intense and relevance signals are commoditised; requires exceptional depth, original research, or unique angle to break through.
- Content Decay
- The tendency for page traffic to decline over time as competition increases, information becomes outdated, or search intent shifts. Addressed through content refreshes or updates.
- Context Window
- The maximum length of text (measured in tokens) that a language model can process in a single request. Larger context windows allow LLMs to consider more information when generating responses.
D
- DA (Domain Authority)↗
- Moz's proprietary score (0-100) estimating a domain's link-based authority. Not used by Google directly.
- Disavow file↗
- A file uploaded via Search Console listing backlinks Google should ignore. Useful for manual action recovery; widely overused otherwise.
- Dofollow
- A link without rel="nofollow". Passes link equity normally. The default state of any link.
- DR (Domain Rating)↗
- Ahrefs' proprietary score (0-100) for a domain's link profile strength. Logarithmic; not used by Google directly.
- Digital PR↗
- An off-page SEO tactic that earns editorial backlinks and brand coverage by creating and pitching newsworthy content, data, or stories to journalists and publishers.
- Duplicate content↗
- Substantially identical or very similar content appearing at multiple URLs. Dilutes ranking signals across pages; resolved with canonical tags, redirects, or consolidation.
E
- E-E-A-T↗
- Experience, Expertise, Authoritativeness, Trustworthiness. Google's quality framework, particularly stringent for YMYL content.
- Entity SEO↗
- Optimising for search engines and AI systems that reason about real-world entities (people, organisations, products) rather than keywords.
- Editorial Calendar
- A publishing schedule that tracks content assignments, deadlines, and publication dates. Essential for maintaining consistent topical coverage and distributing work across a team.
- Editorial Link
- A backlink earned without solicitation, placed by another site owner or editor because they found the content valuable and relevant. The highest-quality link type because it indicates genuine endorsement.
- Embedding
- A numerical vector representation of content (word, passage, or document) that captures its semantic meaning. Used by language models and search systems to calculate similarity and relevance.
- Engagement rate
- A Google Analytics 4 metric: the percentage of sessions that were engaged sessions. A session is engaged if it lasted longer than 10 seconds, included a conversion event, or had two or more pageviews. Replaced bounce rate as the primary engagement metric in GA4.
F
- Featured snippet↗
- A search result format where Google extracts and displays a passage from a ranking page directly in the SERP, above the standard organic results.
- Faceted navigation
- A filtering system common on e-commerce and catalogue sites that lets users narrow results by attribute: size, colour, brand, price. Can generate large numbers of thin, near-duplicate URL combinations that dilute crawl budget if not controlled with noindex or canonical tags.
- FID (First Input Delay)
- A former Core Web Vital measuring responsiveness to the first user interaction. Replaced by INP in March 2024.
- First-Hand Experience
- Personal, direct involvement with a topic: using a product, living through an event, practising a skill. Signals genuine authority and is now explicitly part of Google's E-E-A-T framework (the first E).
G
- GBP (Google Business Profile)↗
- Google's free local business listing, the primary asset for local pack and Google Maps visibility. Formerly Google My Business.
- GEO (Generative Engine Optimisation)↗
- The practice of structuring content for retrieval and citation by AI answer engines. Overlaps with traditional SEO; emphasises entity clarity and direct-answer formats.
- Googlebot↗
- Google's web crawler. Operates in two-pass indexing: initial HTML processing, then JavaScript rendering.
- Google-Extended↗
- Google's user agent for AI training data collection, separate from Googlebot. Can be blocked in robots.txt without affecting search indexing.
- Google Lens↗
- Google's visual search product. Searches by pointing a camera at a subject, uploading an image, or selecting content on screen. Can identify objects, read text in images, surface similar products, and return local business information. Integrated into the Google app, Google Images, and AI Mode.
- Geotargeting
- Configuring a site or section of a site to signal relevance to users in a specific country or region. Signals include hreflang, ccTLDs, Google Search Console geotargeting settings, and on-page language cues.
- GSC (Google Search Console)↗
- Google's free webmaster tool providing data on indexing status, search performance, crawl errors, and manual actions. The primary diagnostic interface between site owners and Google.
H
- Heading tags↗
- HTML elements (H1 through H6) marking the structural hierarchy of page content. The H1 signals the primary topic; lower levels organise sub-sections. A modest on-page ranking signal.
- Hreflang↗
- An HTML attribute (or sitemap element) declaring the language and regional targeting of a page. Manages multilingual and multi-regional sites.
- HSTS (HTTP Strict Transport Security)↗
- An HTTP header that forces browsers to use HTTPS for the domain. Strong commitment; cannot be quickly reversed.
- Hallucination
- A statement made by a language model that is not supported by the sources it was supposed to ground in. A reliability risk for AI answer engines; grounded answers link back to sources to prevent it.
- Head Term
- A broad, high-volume keyword with generic intent: 'shoes', 'laptop', 'seo'. Opposite of long-tail; more competitive but often targets larger search volumes.
I
- Indexing↗
- The process by which a search engine stores and organises crawled pages, making them eligible to appear in search results.
- INP (Interaction to Next Paint)↗
- A Core Web Vitals metric measuring overall page responsiveness to user interactions. Good: under 200ms.
- Impressions
- The number of times a URL appeared in search results for a given query, regardless of whether it received a click. Reported in Google Search Console; used alongside CTR and average position to assess SERP visibility.
- Index bloat
- An excess of low-value, thin, or duplicate pages in a search engine's index. Wastes crawl budget and dilutes quality signals. Addressed through noindex directives, canonicalisation, and content consolidation.
- Internal link↗
- A hyperlink from one page on a site to another page on the same site. Distributes PageRank and signals topical structure.
J
- JavaScript SEO↗
- The discipline of ensuring search engines can crawl, render, and index content delivered via JavaScript.
- JSON-LD↗
- JavaScript Object Notation for Linked Data. The recommended format for embedding structured data in a webpage as an inline script block, preferred by Google over Microdata and RDFa.
K
- KD (Keyword Difficulty)↗
- A score (0-100) estimating how hard it is to rank for a query, typically based on the link profiles of currently ranking pages.
- Keyword cannibalisation↗
- When two or more pages on the same site target the same keyword or query, causing them to compete against each other in search results. Splits ranking signals and can suppress both pages. Resolved through consolidation, canonical tags, or differentiated targeting.
- Keyword mapping↗
- The process of assigning specific target keywords to individual pages across a site. Prevents cannibalisation and ensures each page has a clear primary search purpose.
- Keyword stuffing↗
- Repetitive, unnatural use of target keywords in content. Damages readability; flagged by quality systems.
- Keyword universe↗
- The complete set of search queries relevant to a business, assembled from multiple discovery sources before any filtering or prioritisation. Built from seed terms, keyword tools, autocomplete, People Also Ask, GSC queries, competitor rankings, and customer language. Treated as a persistent asset that feeds downstream keyword research activities.
- Knowledge graph↗
- A database of entities and their relationships used by Google (and AI systems) to answer entity-based queries.
- Knowledge panel↗
- An information box about an entity (person, organisation, place) shown alongside search results, drawn from Google's knowledge graph.
L
- LCP (Largest Contentful Paint)↗
- A Core Web Vitals metric measuring perceived load speed via the largest visible element. Good: under 2.5 seconds.
- Link equity (PageRank)↗
- The authority a page passes through its outbound links. The basis of Google's original ranking algorithm; still influential.
- Link velocity
- The rate at which a site acquires new backlinks over time. Sudden spikes can appear unnatural and attract algorithmic or manual scrutiny.
- llms.txt↗
- A proposed Markdown file at the root of a domain providing an AI-readable index of a site's most important content. Not yet a formal standard.
- Log file analysis↗
- The process of parsing server access logs to understand exactly which URLs search engine bots crawled, at what frequency, and which were skipped. More reliable than crawl simulation tools.
- LocalBusiness schema↗
- Schema.org structured data declaring a business's local entity properties: name, address, phone, hours, services.
- Local pack
- The map and three business listings shown at the top of search results for queries with local intent.
- Link reclamation↗
- The process of finding and recovering lost backlinks, broken inbound links, or unlinked brand mentions and converting them into active links. Lower effort than new link building.
- Long-tail keyword↗
- A specific, often longer search query with low individual volume but typically clearer intent and lower competition.
- LLM (Large Language Model)
- A neural network trained on large amounts of text data to predict and generate text. Powers AI answer engines like ChatGPT, Claude, and Perplexity. Operates on learned patterns rather than external knowledge retrieval alone.
- LLMO (Large Language Model Optimisation)↗
- A term for the practice of optimising content to be cited by large language models. Emphasises influencing what models know rather than what they retrieve in real time; in practice the content tactics are nearly identical to GEO.
- LSI Keywords
- Latent Semantic Indexing keywords: terms semantically related to a primary keyword that signal topic comprehensiveness. 'Espresso', 'cappuccino', and 'latte' are LSI keywords for 'coffee'; using them can strengthen relevance signals.
M
- Manual action↗
- A penalty issued by Google's spam team after human review. Reported in Search Console; requires reconsideration request to lift.
- MCP (Model Context Protocol)↗
- An open standard developed by Anthropic for connecting AI models to external tools and data sources. Adopted across the AI industry including by Google and Microsoft. WebMCP is the browser-native adaptation that lets websites expose callable tools to in-browser AI agents.
- Meta description↗
- An HTML attribute summarising a page for search engines and link previews. Not a direct ranking factor; influences CTR.
- Mobile-first indexing↗
- Google's policy of using the mobile version of a site as the primary version for indexing and ranking. Standard since 2019.
- Mojibake
- Text corrupted by character encoding mismatches, typically appearing as garbled symbols (e.g. â€" instead of an em dash).
- Multimodal search↗
- Search that combines more than one input type in a single query: image and text, video and voice, or camera and spoken question. Google Lens and AI Mode both support multimodal queries; the visual input narrows the query beyond what text alone can express. Multimodal queries are more likely to trigger AI Overviews than equivalent text-only queries.
- MSV (Monthly Search Volume)↗
- The estimated number of times a keyword is searched per month within a given market. A primary input in keyword research alongside keyword difficulty.
N
- NAP↗
- Name, Address, Phone. The core business information that must be consistent across local citations for local SEO.
- Noindex↗
- A meta tag (or HTTP header) instructing search engines not to index a page. Pages must be crawlable for the directive to be seen.
- Nofollow↗
- A link attribute (`rel="nofollow"`) telling search engines not to pass link equity through the link. Used for sponsored, user-generated, or untrusted links.
O
- Off-page SEO↗
- Optimisation activities that happen outside a site: link building, digital PR, brand mentions, citations.
- On-page SEO↗
- Optimisation of elements on individual pages: title tags, headings, content, internal linking, schema.
- Open Graph↗
- Meta tags (og:title, og:description, og:image) controlling how a URL appears when shared on social media and messaging platforms.
- Organic search↗
- Search results not paid for by advertising. Organic traffic is the audience SEO is built around.
- Orphan page
- A page with no internal links pointing to it. Difficult for crawlers to discover and receives no internal link equity.
- Outbound link
- A hyperlink from a page on your site to a page on a different domain. Passes link equity if dofollow, and signals topical associations to search engines.
P
- PageRank↗
- Google's original algorithm for valuing pages based on the quantity and quality of inbound links. Still influential, though heavily evolved.
- Pagination↗
- Splitting long content or lists across multiple pages. Affects crawl, indexing, and how authority flows through a site.
- Passage indexing↗
- Google's ability to index and rank individual passages within a page independently of the page's overall topic, allowing relevant sections to surface for specific queries.
- PBN (Private Blog Network)↗
- A network of sites created specifically to manipulate link equity. Violates Google's guidelines.
- Pillar page↗
- A comprehensive overview page on a topic, linking to deeper cluster pages on specific sub-topics.
- People Also Ask (PAA)↗
- An expandable SERP feature showing related questions users commonly ask for a query. Expanding PAA reveals answers; answering these sub-questions in content can improve visibility and CTR.
- Personalization
- The adjustment of search results based on a user's location, search history, device type, and other signals. Means the same query can produce different results for different users, complicating rank tracking.
- Prompt Engineering
- The practice of designing input prompts to guide language models toward specific, accurate outputs. Includes techniques like chain-of-thought reasoning and few-shot examples.
Q
- Query
- The words or phrase a user types into a search engine. Distinct from 'keyword' in that a query is what the user typed; a keyword is what the SEO targets.
- Query fan-out↗
- A technique used by Google AI Mode in which a single user query is expanded into multiple parallel sub-queries, each targeting a different aspect of the question, with results synthesised into a single answer. Enables AI Mode to retrieve broader and deeper information than a conventional single-pass search. In multimodal queries, AI Mode fans out across both the scene as a whole and individual objects within an image.
R
- Redirect (301)↗
- An HTTP status code indicating a permanent URL change. Transfers ranking signals to the destination URL.
- Redirect (302)↗
- An HTTP status code indicating a temporary URL change. Does not transfer ranking signals.
- Redirect chain↗
- A sequence of redirects (URL A → URL B → URL C). Reduces signal transfer and adds latency. Best avoided.
- Rendering↗
- The process of executing JavaScript and producing the final DOM. Googlebot renders pages in a separate pass after initial HTML processing.
- Rich result
- An enhanced search result with visual elements (star ratings, FAQ accordions, recipe cards, product information) driven by structured data.
- Robots.txt↗
- A plain text file at the root of a domain declaring which crawlers can access which paths. A request, not enforcement.
- RankBrain
- Google's machine learning ranking component (introduced 2015) that learns from user behaviour and search patterns. Treats similar queries as equivalent and predicts relevance for never-before-seen queries.
- Referring Domain
- A unique domain that links to your site. A site earning backlinks from 10 different domains is stronger than one with 100 links from the same domain; domain diversity matters more than link count.
- Retrieval-Augmented Generation (RAG)↗
- A technique that combines document retrieval with language model generation: the system retrieves relevant passages, then feeds them to an LLM for synthesis. Allows grounding responses in current information rather than parametric knowledge alone.
S
- Schema markup↗
- Structured data using the schema.org vocabulary, declaring the meaning of page content explicitly to search engines and AI systems.
- Search intent↗
- The underlying goal behind a query: informational, navigational, commercial investigation, or transactional.
- Search volume↗
- See MSV (Monthly Search Volume).
- SEO (Search Engine Optimisation)↗
- The practice of improving a site's visibility in unpaid search results.
- SEO audit↗
- A structured assessment of a website's technical health, on-page quality, and off-page authority, used to identify issues and prioritise improvements. Can cover the full site or focus on a specific area such as technical SEO or content.
- SERP (Search Engine Results Page)
- The page returned in response to a search query, including organic results, ads, and SERP features.
- SERP features↗
- Non-standard result formats displayed alongside or in place of regular organic listings: featured snippets, AI Overviews, local packs, knowledge panels, image carousels, and more. Triggered by query type and structured data.
- Share of voice
- A site's proportion of total visible traffic or impressions for a defined set of target keywords, relative to competitors. Used as a market-level measure of SEO performance over time.
- Site migration↗
- A significant structural change to a website: moving to a new domain, switching platforms, changing URL structures, or consolidating sites. Carries high risk of ranking loss if redirects, canonical signals, and crawl directives are not handled correctly.
- Sitemap↗
- An XML file listing the URLs of a site, providing search engines with a catalogue for crawling.
- Sitelinks
- Additional links to sub-pages shown beneath a brand's main search result. Algorithmically generated; not directly controllable but influenced by site structure and internal linking.
- Soft 404
- A page returning a 200 OK status code but containing content that signals it should be a 404. Confuses indexing.
- Structured data↗
- Machine-readable metadata embedded in a webpage. Most commonly delivered as JSON-LD using schema.org vocabulary.
- Search Intent Drift
- A shift in the dominant intent of a keyword over time, often from informational to commercial or vice versa. Requires content format and angle adjustments; a page optimised for the old intent may lose rankings as user needs evolve.
- Seed Keyword
- A broad head term or topic that starts keyword research. 'Keyword research', 'SEO', 'content marketing' are seed keywords; cluster research expands them into related long-tail terms.
- Server-Side Rendering (SSR)
- Rendering HTML on a web server before sending it to the browser. Full content is delivered to crawlers immediately; contrasts with Client-Side Rendering where crawlers may see incomplete pages.
T
- Thin content↗
- Pages with little or no unique value: doorway pages, scraped content, auto-generated text, or pages with minimal original material. A quality signal targeted since the Panda era.
- Title tag↗
- The HTML element (`<title>`) defining a page's title. Appears in browser tabs, SERPs, and link previews. One of the most influential on-page ranking factors.
- Topic cluster↗
- A group of related pages organised around a pillar page, used to build topical authority.
- Topical authority↗
- A measure of how comprehensively and credibly a site covers a subject area, built through depth of content and internal linking structure. Influences rankings across a topic regardless of individual page signals.
- Token
- The basic unit of text processed by language models: typically a word or subword. Context window limits are measured in tokens, not characters or words. Longer responses consume more tokens.
U
- UGC (rel=ugc)↗
- A link attribute (`rel="ugc"`) indicating the link was created by a user, such as in comments or forum posts. Introduced alongside rel=sponsored in 2019 as a refinement of nofollow.
- URL (Uniform Resource Locator)↗
- The address of a webpage. Clean, descriptive URLs reinforce topical relevance and improve usability.
- Unlinked Mention↗
- A reference to a brand, product, or topic without a hyperlink. Unlinked mentions signal brand authority to search engines; link reclamation converts them into actual backlinks.
- User Agent
- A string of text sent by a browser or crawler with each request, identifying itself and its capabilities. Google operates multiple user agents (Googlebot, Google-Extended); different agents receive different treatment.
V
- Viewport meta tag
- An HTML meta tag (`<meta name="viewport">`) controlling how a page scales on mobile devices. Essential for passing mobile-friendliness signals.
- Voice search↗
- Searches performed by speaking to a device rather than typing. Queries are longer and more conversational than typed searches, with disproportionate local intent. Spoken results are sourced predominantly from the featured snippet: only one answer is read aloud, making it a winner-takes-all surface.