This SEO glossary explains the terms used in search, website optimisation and AI visibility, with practical distinctions that help you use them correctly. Crawling, indexing, ranking and being cited in an AI answer are different processes, so success in one does not guarantee the next.
Browse the categories below or use your browser’s Find function to jump to a term. For a practical sequence of checks, use our SEO checklist.
Key Takeaways
- Crawling, indexing, ranking and AI citations are separate processes with different checks.
- AI monitoring scores depend on the prompts, platforms and counting rules used. They do not automatically measure market share.
- SEO terminology combines official product definitions, third-party metrics and industry shorthand. Check which one a claim refers to.
Search Engine Terms
Search engines discover information, organise it and return results for a query. These terms describe the products and processes involved.
Algorithm. A set of rules or computational steps used to solve a problem. Search engines combine many algorithms and systems to interpret queries and select results.
Baidu. A search engine and technology company with a substantial presence in China. Its products, results and webmaster requirements differ from Google’s.
Bing. Microsoft’s search engine. Bing Webmaster Tools provides site owners with information about crawling, indexing and search performance in Bing.
Crawling. Fetching a URL’s content using software such as Googlebot. A crawler can visit a page without that page being indexed or appearing in results.
Discovery. Finding that a URL exists, for example through a hyperlink or sitemap. Discovery can happen before a search engine decides whether to fetch the URL.
DuckDuckGo. A search engine that emphasises privacy. It combines information from several sources, including its own crawler and search partners.
Featured Snippet. A search result that highlights an extract answering a query. It normally attributes the extract to a source page and links to it.
Google Ads. Google’s advertising platform, formerly called Google AdWords. Paying for search ads does not buy higher placement in organic results.
Google Business Profile (GBP). A free tool for eligible businesses to manage information shown on Google Search and Maps. It was previously called Google My Business.
Google Lens. Google’s visual search tool. It uses an image or camera input to help identify objects, find related information or search for products.
Google Search Console (GSC). Google’s site-owner tool for search performance, indexing diagnostics and related reports. It measures activity in Google Search rather than every visit to your website.
Googlebot. Google’s main web crawler, with smartphone and desktop variants. Server logs can show requests, but a user-agent name alone does not prove a request came from Google.
Indexing. Analysing content and storing information about it in a search engine’s index. Google’s explanation of Search distinguishes this from crawling and serving results.
Knowledge Graph. A structured representation of entities and their relationships. For example, a graph can connect a business with its founder, location and services.
Knowledge Panel. An information panel about an entity in search results. A Knowledge Graph is the underlying information structure, while a panel is a visible presentation.
Local Pack. A group of local business results, often accompanied by a map. Its composition can vary with the query and the searcher’s location.
Local SEO. Improving a business’s visibility for searches with local intent. Local SEO work can involve business information, relevant pages, reviews and local links.
Naver. A South Korean search platform that combines web results with its own content and services. Research its actual results when targeting Korean audiences.
Organic Result. An unpaid search listing. Organic results can include links to pages, images, videos and other content formats.
People Also Ask (PAA). A Google results feature containing related questions and expandable answers. The displayed questions can provide research ideas, but they are not search-volume measurements.
Personalisation. Adjustments to results based on information about a user. Location, language and device can also change results without relying on personal search history.
Query. The input submitted to a search system. A query can contain words, an image, voice input or a combination, depending on the product.
Ranking. The ordering of results for a particular search. Positions depend on the query, result type, location, device and time of measurement.
Ranking Signal. Information a search system uses when evaluating results. A third-party tool score should not be assumed to be a signal used by Google.
Rich Result. A search result with additional supported features, such as product information or recipe details. Valid structured data can establish eligibility without guaranteeing display.
Search Engine Marketing (SEM). Marketing through search engines. Some teams use SEM for paid search specifically; others include both SEO and paid search, so define the scope.
Search Engine Optimisation (SEO). Improving how people and search engines find, understand and use a website’s content through unpaid search. It covers content, technical implementation and reputation.
Search Engine Results Page (SERP). The page or interface displaying results for a search. Modern SERPs can combine organic listings, ads, local results and AI features.
SERP Feature. A distinct result format, such as a local pack, video carousel or featured snippet. Feature availability affects how people encounter and interact with results.
Sitelinks. Additional links beneath a main search result that help users reach sections or pages within a site. Google usually selects these automatically.
Social Search. Searching within platforms such as YouTube, TikTok or Reddit. Our social search guide explains how these searches can fit into a wider research journey.
Vertical Search. Search focused on a particular subject or format, such as products, jobs, images or flights.
Web Crawler. Software that requests web resources automatically. Crawlers serve different purposes, including search discovery, audits and data collection.
Yandex. A search engine and technology company associated with the Russian-language market. Its webmaster tools and search behaviour require separate assessment from Google.
Algorithm Terms
An update changes a search system; a system can continue operating between announced updates. Historical update names are useful context, but they are not a current optimisation checklist.
BERT. A language-understanding model used in Google Search to interpret relationships between words and context. It does not create a special list of keywords to add.
Core Update. A broad change to Google’s core ranking systems. A traffic decline during an update requires investigation; timing alone does not identify the cause.
Deduplication. Selecting representative results when multiple pages contain substantially similar information. This reduces repetition in search results.
Freshness Systems. Systems that favour recent information when a query calls for it. Changing a publication date does not make unchanged information current.
Helpful Content System. Google’s former standalone helpful-content system. Its work became part of core ranking systems in March 2024, as documented in the ranking systems guide.
Hummingbird. Google’s name for a substantial overhaul of its search systems in 2013. It is a historical reference, rather than a separate current ranking target.
Manual Action. An action applied by a human reviewer when content violates Google’s spam policies. Search Console reports manual actions; ordinary ranking losses are not automatically manual actions.
Panda. A historical Google system focused on content quality. It later became part of Google’s core ranking systems.
Passage Ranking. Google’s use of individual sections to understand a page’s relevance. The page remains the indexed result; its paragraphs do not become separately indexed pages.
Penguin. A historical Google system addressing link spam, later incorporated into core ranking systems. Its name still appears in older SEO advice.
RankBrain. A Google system that connects words with concepts to help interpret searches and relevant content, including where wording does not match exactly.
Ranking Volatility. Changes in observed search positions over time. Updates, competitors, seasonality, technical faults and measurement differences can all contribute.
Relevance. How closely content addresses the meaning and requirements of a query. Mentioning a word is insufficient if the page does not satisfy the task.
Reviews System. Google’s system for evaluating review content, including the quality of analysis and original research. It is distinct from a business’s star rating.
Spam Update. An announced improvement to Google’s spam-detection systems. It is different from a manual action on a specific site.
SpamBrain. Google’s AI-based spam-detection system. It helps identify content and behaviour that violate search spam policies.
Quality Rater Terms
Google’s quality raters evaluate examples of search results to help assess its systems. Their ratings do not directly set an individual page’s ranking.
Authoritativeness. Recognition that a source or creator is a reliable reference for a subject. It depends on the topic and evidence, rather than a universal website score.
Beneficial Purpose. A page’s intended helpful purpose, such as explaining a task, selling a product or entertaining an audience. Its quality is assessed in that context.
E-A-T. The older abbreviation for expertise, authoritativeness and trustworthiness. Google later added experience to form E-E-A-T.
E-E-A-T. Experience, expertise, authoritativeness and trustworthiness. It is a framework discussed in Google’s content guidance, not a single ranking factor or publicly available score.
Experience. First-hand involvement with a subject, such as actually using a product or completing a process. Describe the evidence rather than simply claiming experience.
Expertise. The knowledge or skill appropriate to a topic. The level required depends on the consequences of giving incorrect advice.
Main Content. The part of a page that directly serves its purpose, such as the article text, product details or working calculator.
Needs Met. A quality-rating assessment of how well a result satisfies a particular query in its context.
Page Quality. A quality-rating assessment of how well a page achieves its purpose, considering its content, creator, reputation and other relevant evidence.
Supplementary Content. Material supporting the experience without being the main content, such as navigation and related links.
Trustworthiness. The extent to which information and its source are accurate, honest, safe and reliable. HTTPS alone cannot establish that a claim is true.
YMYL. Your Money or Your Life. Topics where inaccurate information could significantly affect health, financial stability, safety or wider societal wellbeing.
AI SEO Terms
AI visibility involves several different stages, from retrieving information to generating an answer and displaying sources. These terms help separate what a system does from what a monitoring tool can actually observe.
AI Agent. A system that uses a model and tools to carry out a task through multiple steps. Its capabilities depend on its tools, instructions and permissions.
Agentic Search. Search in which an AI system plans and performs several retrieval steps to work towards a user’s task, potentially refining its approach along the way.
AI Mode. Google’s conversational search experience for questions requiring exploration, reasoning or comparisons. It can generate responses with links to supporting websites.
AI Overviews (AIO). Google’s AI-generated summaries within Search. They are separate from featured snippets and are not shown for every query.
AI Overview Coverage. The proportion of a defined keyword sample whose observed results contain an AI Overview. Record the market, device, date and number of keywords tested.
AI SEO. SEO work concerned with AI-powered search experiences, as well as the use of AI in SEO workflows. State which meaning applies when comparing services or results.
AI Visibility. How often and in what context a brand or source appears in observed AI answers. Our AI visibility tools guide covers differences between monitoring approaches.
Answer Engine. A system that responds with an answer assembled or generated from information sources. Whether it provides citations or performs live retrieval depends on the product and request.
Answer Engine Optimisation (AEO). Work intended to improve a source’s usefulness and visibility in direct answers, including AI responses and other answer formats. The term has no universal technical specification.
Answer Variability. Differences between responses to the same or similar prompts. Model changes, retrieval, location, conversation context and generation settings can affect the output.
Benchmark Prompt Set. A fixed collection of prompts used for repeatable comparison. Keep prompts and test conditions consistent when interpreting changes over time.
Brand Mention. A reference to a brand in an answer, with or without a link. A mention can be favourable, neutral or negative and may contain inaccuracies.
ChatGPT Search. ChatGPT’s web-search capability, which can retrieve information and provide linked sources. A response generated without search should not be treated as evidence of live web retrieval.
ChatGPT-User. A user agent used for certain user-initiated visits from ChatGPT. It is distinct from OpenAI’s automatic search and training crawlers.
Chunk. A portion of a document processed as a unit for retrieval or model input. A chunk may contain a paragraph, several sections or another division chosen by the system.
Chunking. Dividing documents into manageable pieces. Microsoft’s chunking guidance explains why useful boundaries and retained context affect retrieval.
Citation. A reference attaching an answer or claim to a source. Check the source itself, because a displayed citation can fail to support the associated claim.
Citation Coverage. The percentage of sampled answers that cite a specified domain or URL at least once. Divide answers with that citation by all eligible answers in the sample.
Citation Share. A domain’s citations divided by all citations counted in a defined answer sample. Specify whether repeated links count and which competitors or sources enter the denominator.
Citation Accuracy. Whether a cited source supports the claim attached to it. Source visibility and source support are separate checks.
Context Window. The amount of information a model can process within a request, measured in tokens. Limits and handling of input and output vary by model.
Conversational Search. Search that uses follow-up questions and dialogue context. Later answers can depend on earlier messages rather than the latest query alone.
Corpus. The collection of documents or other material available to a search, retrieval or training system. Different systems can operate on different corpora.
Embedding. A numerical representation of text, images or other data. Embeddings allow systems to compare items using a mathematical distance or similarity measure.
Entity. A distinct person, organisation, place, product or concept. Naming the specific entity can resolve ambiguity that a generic phrase leaves open.
Entity Resolution. Determining when different names or references identify the same entity. Consistent business details can make this easier without guaranteeing a particular search result.
Fine-Tuning. Additional training that adjusts an existing model for a task or behaviour. Updating a website does not directly fine-tune a public AI model.
Foundation Model. A broadly trained model that can support a range of downstream tasks. It may be adapted through instructions, tools or further training.
Generative AI. AI that produces content such as text, images, audio or code. Generated output can be plausible while still containing errors.
Generative Engine Optimisation (GEO). Improving content and supporting evidence for discovery, use and citation in generative search. Our GEO guide explains the practical work involved.
Google-Extended. A robots.txt product token controlling certain uses of crawled content for Gemini training and grounding. It is not a separate HTTP crawler and does not control Google Search inclusion.
GPTBot. OpenAI’s crawler for content that may be used to train its foundation models. Its robots.txt control is independent of OAI-SearchBot.
Grounding. Providing external information to support a model’s response. Grounding can reduce unsupported output, but does not guarantee that the model interprets or cites sources correctly.
Hallucination. Model output that invents information or presents unsupported claims as factual. Examples include nonexistent sources, incorrect figures and fabricated quotations.
Hybrid Search. Retrieval that combines methods such as keyword and vector search. Microsoft’s hybrid-search documentation shows how result sets can be combined.
Inference. Using a trained model to produce output from an input. Retrieving a web page during inference is different from adding that page to training data.
Large Language Model (LLM). A model trained on large quantities of language data to process and generate text. Tools can extend its access to current information and external systems.
Lexical Search. Retrieval based on words and their indexed representations. It is useful for precise names, product codes and phrases where exact wording carries meaning.
llms.txt. An optional, proposed Markdown format for introducing a website and linking to useful material for AI systems. The llms.txt proposal is not a replacement for robots.txt, an access-control mechanism or a guarantee of citation.
Mention Rate. The percentage of sampled answers that mention a brand. For example, a brand mentioned in 12 of 40 answers gives a 30% answer-level mention rate, not 30% of the market.
Model Training. The process of adjusting a model’s parameters using data. Training and live search retrieval have different purposes, timing and publisher controls.
Multimodal Model. A model that processes or generates more than one type of input or output, such as text and images. Supported combinations depend on the model.
OAI-SearchBot. OpenAI’s crawler for surfacing sites in ChatGPT search features. OpenAI documents its crawler controls separately from GPTBot and user-initiated requests.
Prompt. Input provided to an AI system, including a question, task or instruction. The complete model input can also include conversation history and system instructions.
Prompt Injection. Instructions inside untrusted material that attempt to redirect an AI system’s behaviour. Text on a retrieved page should be treated as source content, not automatic authority over the task.
Prompt Research. Investigating the questions, context and constraints people use in AI-assisted research. A generated prompt list is a hypothesis set unless supported by observed customer behaviour or other evidence.
Query Fan-Out. Issuing several related searches to help answer one request. Google says its AI search features may use this technique across subtopics and sources.
Query Rewriting. Reformulating a request for a retrieval system, for example by resolving a pronoun from earlier conversation. The retrieved query may differ from the user’s wording.
Retrieval. Selecting relevant information from an available collection or source. Being retrieved does not necessarily mean being cited in the final answer.
Retrieval-Augmented Generation (RAG). Combining retrieved information with model generation. Microsoft’s RAG overview describes how retrieval supplies material for an answer without retraining the model for each update.
Retrieval Precision. The proportion of retrieved items that are relevant in an evaluation. It requires a defined relevance judgement, not just a count of returned documents.
Retrieval Recall. The proportion of all relevant items that the system retrieves in an evaluation. Public citations alone do not reveal everything the system retrieved or missed.
Reranking. Reordering an initial set of retrieved candidates using another relevance assessment. It refines the shortlist rather than searching the whole corpus again.
Semantic Search. Search that considers meaning and context beyond literal word matches. Vector retrieval and semantic reranking are possible components, rather than interchangeable names for the whole process.
Semantic Similarity. A measure of how closely items relate in meaning under a particular model. A high score does not establish factual agreement or truth.
Search Generative AI Control. A Search Console setting controlling inclusion in Google’s specified generative search experiences. Google’s control documentation distinguishes this display and grounding setting from model-training controls.
Sentiment. An assessment of favourable, neutral or unfavourable language about a subject. Automated scores need checking, especially for comparisons, sarcasm and mixed evaluations.
Source Attribution. Identifying where information came from. Attribution can be shown as a link, citation marker or named source, and is separate from proof that the information is correct.
Synthetic Prompt. A prompt created for research or testing rather than taken from an observed user request. Useful for testing scenarios, but it has no measured demand by default.
Token. A unit a model uses to represent input or output. Tokens can be words, parts of words or other symbols; one token does not equal one word.
Top-K Retrieval. Returning a selected number of the highest-ranked retrieval candidates. The value of K controls shortlist size, not the number of sources ultimately cited.
Vector Database. A system designed to store and search vectors alongside associated data. It can support retrieval applications, but storing content there does not make it available to public search engines.
Vector Search. Finding items whose vectors are close under a chosen similarity measure. Microsoft’s vector-search overview describes its use in matching related content.
Visibility Share. A comparison of appearances within a specified monitoring sample. The score depends on the tool, prompts, brands and counting rules; it is not automatically market share.
Keyword Terms
Keyword research connects customer language with useful pages. Search-volume estimates help prioritise work, but they do not measure every question or determine whether a page deserves to exist.
Branded Keyword. A query containing a specified brand or a recognised variation. Define whether product brands and competitor brands are included in your reporting rule.
Commercial Intent. Research aimed at evaluating a purchase, such as comparing products, prices or providers. The searcher may still need evidence before choosing.
Head Term. A broad keyword with relatively high demand within a topic. It often requires more context before the desired result is clear.
Informational Intent. A search intended to learn or solve a problem. Informational searches can occur before or after a purchase.
Keyword. A word or phrase used to research and organise search demand. A single useful page can satisfy many related queries without repeating every variation verbatim.
Keyword Cannibalisation. A situation where overlapping pages undermine the site’s ability to satisfy an intended search task. Two pages receiving impressions for the same query do not, by themselves, prove a problem.
Keyword Cluster. Related queries grouped because they share a topic or search task. Check actual results and intent before assuming the whole cluster belongs on one page.
Keyword Competition. In Google Keyword Planner, the level of advertiser competition. It measures paid-search participation and should not be treated as an organic ranking difficulty score.
Keyword Density. The frequency of a keyword relative to the amount of text. There is no universal percentage that makes a page rank well.
Keyword Difficulty (KD). A tool’s estimate of the difficulty of ranking for a query. Providers use different methods, and the score does not replace inspecting the results.
Keyword Gap. Relevant demand covered by competing sites but missing or poorly served on your own. A competitor’s ranking alone does not make the topic suitable for your business.
Keyword Mapping. Assigning query groups and search tasks to the most appropriate URLs. The map helps identify missing content, overlaps and pages needing a clearer purpose.
Keyword Modifier. A word or phrase narrowing a search, such as a location, use case, price or product attribute.
Keyword Research. Investigating what people search for, what they need and which pages can serve them. Our keyword and AI prompt research guide covers the workflow.
Keyword Stuffing. Unnatural repetition or insertion of keywords to manipulate rankings. Google’s spam policies cover it; listing every suburb or phrase variation is not useful optimisation.
Long-Tail Keyword. A query in the low-demand part of the search distribution, often expressing a specific need. Long-tail status is about frequency, not a fixed minimum word count.
LSI Keywords. An SEO label often misapplied to related words. Latent semantic indexing is an information-retrieval technique, not a special category of keywords Google requires writers to add.
Navigational Intent. A search intended to reach a particular website, page or product, such as a brand’s customer login.
Non-Branded Keyword. A query that does not include the brand terms specified in your reporting rules. Classification depends on those rules, not a universal keyword property.
Search Intent. The task behind a search. Informational, navigational, commercial and transactional labels are useful starting points, but a query can support more than one task.
Search Volume. An estimate of searches for a keyword within a stated market and period. It is not a count of unique people, available clicks or likely customers.
Seasonality. Predictable changes in demand across a year or recurring cycle. Compare equivalent periods before attributing a seasonal decline to content quality.
Seed Keyword. A starting topic used to discover more detailed searches. A seed such as “heat pump” can lead to questions about costs, sizing and installation.
SERP Intent Analysis. Reviewing actual search results to understand the page types and tasks a search engine currently prioritises for a query.
Transactional Intent. A search aimed at completing an action, such as buying, booking, subscribing or downloading.
Zero-Volume Keyword. A query reported as having no measurable search volume by a tool. It may still reflect a real need; our zero-volume keyword guide covers how to investigate these queries.
Technical SEO Terms
Technical SEO concerns how a website delivers content and how search engines access and interpret it. Diagnose the specific issue before choosing a redirect, canonical, crawl rule or indexing directive.
200 OK. An HTTP status indicating a successful request. A 200 response does not, by itself, mean the page is indexable, indexed or useful.
301 Redirect. An HTTP response indicating a permanent move to another URL. Use a relevant destination when a page is replaced or consolidated.
302 Redirect. An HTTP response indicating a temporary redirect. It communicates a different intention from a permanent move.
404 Not Found. An HTTP status indicating that the requested resource cannot be found. A helpful error page can still correctly return 404.
410 Gone. An HTTP status indicating that a resource has been intentionally removed and is not expected to return.
5xx Error. A class of server-error responses, such as 500 or 503. Persistent failures can prevent users and crawlers from accessing content.
AMP. Accelerated Mobile Pages, an open-source framework for creating pages with constrained components and performance conventions. AMP is not required for ordinary mobile indexing.
Async. A script-loading attribute allowing a script to download without blocking HTML parsing and execute when available. Execution order requires care when scripts depend on one another.
Breadcrumbs. Navigation showing a page’s place in a hierarchy. Breadcrumb links help users move to parent categories and give crawlers additional routes.
Broken Link. A link whose destination does not deliver the intended resource. Check the reason before replacing it, removing it or redirecting a retired page.
Caching. Storing a reusable copy of a resource or response to reduce repeated work. Cached content can become stale if invalidation is not handled correctly.
Canonical Tag. A rel=”canonical” annotation identifying a preferred representative of duplicate or very similar content. It is a signal, not a command; Google can choose another canonical.
Canonical URL. The representative URL selected for a set of duplicate or similar pages. A declared canonical and Google’s selected canonical can differ; our canonical tags guide explains the implementation.
ccTLD. A country-code top-level domain, such as .au or .uk. Registration rules vary by registry; .com.au is a namespace beneath .au.
Client-Side Rendering (CSR). Building or updating page content in the browser with JavaScript. Test whether essential content and links are available to the crawlers you need to serve.
Content Delivery Network (CDN). A distributed network that delivers resources from locations closer to users. Its caching and security rules can also affect crawler access.
Crawl Budget. The amount of crawling Google can and wants to perform on a site. It reflects crawl capacity and demand, rather than a fixed allowance of indexed pages.
Crawl Depth. The number of link steps needed to reach a URL from a chosen starting page. It is different from counting folders in a URL.
Crawl Trap. A structure that generates excessive or effectively unlimited crawlable URLs, such as unrestricted calendar navigation or combinations of filters.
CSS. Cascading Style Sheets, the language used to style web documents. CSS controls presentation; it does not guarantee complete indexing.
Defer. A script-loading attribute that lets an external classic script download during HTML parsing and execute afterwards. Deferred classic scripts preserve their document order.
Deindexing. Removal of a URL’s content from a search engine’s index. Check indexing evidence rather than inferring deindexing from fewer impressions alone.
DNS. Domain Name System, which resolves domain names to information such as server IP addresses. DNS failures can make a site unreachable.
DOM. Document Object Model, the browser’s structured representation of a document. JavaScript can change the DOM after the initial HTML is received.
Duplicate Content. Substantially similar content accessible at multiple URLs. It can complicate canonical selection and crawling, but duplication alone is not automatically a spam violation.
Faceted Navigation. Filters that narrow a collection by attributes such as size, colour or price. Filter combinations need deliberate URL and crawl handling.
Hreflang. An annotation identifying alternate language or regional versions of a page. It helps search engines choose an appropriate version and does not translate content.
HTML. HyperText Markup Language, used to describe the structure and meaning of a web document, including headings, paragraphs and links.
HTTP. Hypertext Transfer Protocol, the protocol used for requests and responses between web clients and servers.
HTTPS. HTTP protected with transport encryption and server authentication using TLS. It protects the connection, but does not verify every claim made on a website.
Indexability. Whether a page appears technically eligible for indexing. Eligibility does not establish that a search engine has chosen to index it.
Internal Link. A link between pages on the same website. Our internal linking guide explains how relevant links support navigation and discovery.
JavaScript SEO. Work ensuring that JavaScript-driven content, links and metadata can be discovered, rendered and processed correctly by search engines.
JSON-LD. A JSON-based format for linked data, commonly used to add structured data to a page. Its claims should match the visible content.
Lazy Loading. Deferring resources until they are needed. Poor implementation can hide content from crawlers or delay the main image users need immediately.
Log File Analysis. Examining server request records to understand access patterns, status codes and crawler activity. Verify bot identities before treating requests as search-engine behaviour.
Mobile-First Indexing. Google’s use of the mobile version of content for indexing. Important information and metadata should remain available on mobile.
Nofollow. A rel=”nofollow” link qualification. Google treats it as a hint; it is not a reliable way to prevent the destination from being crawled or indexed.
Noindex. A directive requesting exclusion from search results. A crawler must be able to access the relevant meta tag or HTTP header to process it.
Orphan Page. A page with no discoverable internal links pointing to it within the site being audited. It may still be found through a sitemap or external link.
Pagination. Dividing a collection across multiple pages with distinct URLs. Our pagination guide covers how links between pages help users and crawlers reach items beyond the first page.
Redirect Chain. A sequence of redirects before the final destination. Linking directly to the final relevant URL avoids unnecessary hops.
Redirect Loop. Redirects that return to an earlier URL without reaching a final page. Loops prevent access until the rules are corrected.
Rendering. Processing page resources to produce the document a browser can display. Rendered content may differ from the initial HTML response.
Robots.txt. A file specifying crawl preferences for compliant bots. Robots.txt controls crawling, and a blocked URL can still appear in results without its content.
Schema.org. A shared vocabulary for describing entities and content in structured data. Individual search engines support only some types for specific search features.
Self-Referencing Canonical. A canonical annotation that points to the page’s own preferred URL. It clarifies the intended representative version without guaranteeing selection.
Server-Side Rendering (SSR). Producing page HTML on the server before delivering it. It can make content available in the initial response, though implementation still needs testing.
Site Architecture. The organisation of pages and the connections between them. Categories, navigation and contextual links all contribute.
Soft 404. A response that appears to describe missing or empty content while returning a success status. Search engines may classify it as an error despite the 200 response.
Structured Data. Machine-readable descriptions of content or entities. Accurate markup can support understanding and feature eligibility, but cannot compensate for missing visible information.
Subdomain. A domain beneath another domain, such as shop.example.com. Its relationship to the main site depends on the implementation and purpose.
Subfolder. A path segment below the domain, such as example.com/guides/. Folders help organise URLs, but structure alone does not establish content quality.
Technical SEO. Improving the site’s delivery, accessibility to crawlers and search interpretation. Technical SEO audits connect identified problems with implementation checks.
URL. Uniform Resource Locator, an address identifying a resource. A web URL can contain a scheme, hostname, path, query parameters and fragment; see our URL best practices for website examples.
URL Parameter. A value in a URL’s query string, such as ?colour=blue. Parameters can change content, track a visit or perform another function.
XML Sitemap. A file listing URLs a site wants search engines to discover, optionally with metadata. Inclusion does not guarantee crawling or indexing.
X-Robots-Tag. An HTTP response header carrying indexing or preview directives. It can apply to resources such as PDFs as well as HTML pages.
Page Experience Terms
Performance measurements describe different parts of an experience. Real-user data and a simulated test can disagree because they measure different visits and conditions.
Accessibility. Designing and building content that people with different abilities can use, including those using assistive technology. Automated checks cover only part of accessibility testing.
Chrome User Experience Report (CrUX). Google’s dataset of real-world experience measurements from eligible Chrome users. Not every page has enough eligible observations for URL-level data.
Core Web Vitals. Three metrics covering loading, responsiveness and visual stability. The current set is LCP, INP and CLS, with recommended thresholds assessed at the 75th percentile of visits.
Cumulative Layout Shift (CLS). A measure of unexpected visual movement. A good CLS value is 0.1 or less at the recommended percentile.
Field Data. Measurements collected from actual users under real conditions. Device mix, networks, caching and user behaviour all influence the observations.
First Input Delay (FID). A former Core Web Vital measuring delay before processing a user’s first interaction. INP replaced FID as a Core Web Vital in March 2024.
Interaction to Next Paint (INP). A measure of page responsiveness based on interactions during a visit. A good INP is 200 milliseconds or less at the recommended percentile.
Lab Data. Measurements from controlled or simulated tests. Useful for diagnosing problems and comparing changes, but not a substitute for actual user experience data.
Largest Contentful Paint (LCP). The time until the largest eligible visible content element renders. A good LCP is 2.5 seconds or less at the recommended percentile.
Lighthouse. An automated tool auditing aspects of performance, accessibility, best practices and SEO. A high score does not prove that a site will rank well.
PageSpeed Insights (PSI). Google’s tool combining Lighthouse diagnostics with available CrUX field data. Check whether the field view represents the specific URL or its origin.
Responsive Design. Layout and content behaviour that adapt to the available screen space. Test menus, forms and tables as well as how the page looks.
Time to First Byte (TTFB). The time between starting a navigation request and receiving the first response byte. It includes network and server-related delays, not just application processing.
User Experience (UX). A person’s overall experience of using a product or page. Useful content, readable presentation and working interactions all contribute.
Content SEO Terms
Useful content satisfies a task with enough detail and evidence. Length, publication frequency and the number of pages are not substitutes for that outcome.
B2B Content. Content intended for people buying or working on behalf of organisations. It may need to address several stakeholders and purchasing requirements.
B2C Content. Content intended for individual consumers. The level of detail still depends on the decision, product and audience.
Content Audit. Reviewing existing pages against their purpose, accuracy, performance and usefulness. An audit can identify material to keep, improve, consolidate or retire.
Content Brief. Instructions for a specific page, including its audience, search task, scope, evidence requirements and links. It should guide a useful answer rather than dictate keyword repetition.
Content Consolidation. Combining overlapping material into a stronger destination. Preserve useful information and choose redirects according to how closely the old and new pages serve the same task.
Content Decay. A decline in a page’s performance over time. Outdated information is one possible cause; changes in demand, competition and result layouts can also contribute.
Content Gap Analysis. Identifying needs that existing content does not adequately serve. Gaps can be missing explanations within a page as well as missing pages.
Content Hub. A central page connecting useful resources on a topic. The hub should help someone navigate the subject rather than simply list every related URL.
Content Pruning. Removing or retiring material that no longer serves a useful purpose. Low traffic alone is insufficient evidence when a page supports customers, links or conversions.
Content Refresh. Updating a page’s substance, accuracy or task coverage. Changing a date or adding words without improving the answer is not a meaningful refresh.
Content Strategy. Deciding which audience needs to serve, what to create or improve, and how to maintain it. SEO content strategy connects these decisions with search evidence.
Evergreen Content. Material whose core subject remains useful over time. Examples, interfaces, links and technical details can still need maintenance.
First-Party Evidence. Information collected directly through your own work, research or systems. State its scope and method so others can judge what it supports.
Information Gain. Additional useful knowledge a page contributes beyond what a reader already has or can find elsewhere. It is an editorial concept here, not a claimed public Google score.
Landing Page. The first page someone reaches in a visit, or a page designed for a particular campaign or action. The meaning depends on the reporting or marketing context.
Pillar Page. A broad resource introducing a substantial topic and connecting relevant supporting pages. It does not need to repeat every detail covered by those pages.
Programmatic SEO. Creating or managing pages using structured data and repeatable templates. Each resulting page still needs accurate information and a distinct useful purpose.
Search-Focused Content. Content designed to satisfy an identifiable search task. It should answer the task naturally, with the format and detail the reader needs.
Thin Content. Material offering little useful substance for its purpose. A short answer can be sufficient; an unnecessarily long page can still provide little value.
Topical Authority. Industry shorthand for recognition and demonstrated expertise across a subject. It is not a publicly available Google score that increases with each new article.
Topic Cluster. A group of related pages with distinct purposes connected through useful links. Closely overlapping questions may fit better as sections of one page.
User-Generated Content (UGC). Material contributed by users, such as reviews, forum posts or comments. Moderation and clear attribution help maintain its usefulness.
On-Page SEO Terms
On-page optimisation aligns visible content and page metadata with the task. Search engines can choose different titles or snippets from those supplied in the HTML.
Alt Text. Text describing an image’s meaning or function for people who cannot see it. Decorative images normally use empty alt text.
Anchor Text. The visible text of a link. A descriptive anchor helps users anticipate the destination and provides context to search engines.
Body Copy. The main written text of a page. It should deliver the answer promised by the heading and search listing.
Call to Action (CTA). A prompt encouraging a relevant next step, such as comparing options, downloading a resource or requesting a quote.
Heading Elements. HTML elements from H1 to H6 that express document hierarchy. Choose headings for structure and meaning, with visual styling handled separately.
H1. A top-level heading identifying the page’s main subject. A clear H1 helps orientation without requiring a particular keyword formula.
H2. A heading for a major section within the page. Use it to separate distinct questions or parts of the task.
H3. A heading for a subsection under an H2. It helps organise detail without turning every short definition into another section.
Meta Description. An HTML summary a search engine may use as a result snippet. Google can instead select text from the page to suit the query.
Open Graph Metadata. Tags describing content for platforms that generate shared-link previews. They are separate from a page’s search title and canonical annotation.
Snippet. The descriptive extract displayed with a search result. Its wording and length can change with the query and presentation.
Title Element. The HTML title naming a document, commonly shown in browser tabs. Google may use it, headings or other sources to create the search result’s title link.
Title Link. The clickable heading of a Google search result. It may differ from the title element supplied by the website.
URL Slug. The readable identifying part of a page’s URL path. Choose a clear, stable description rather than changing it with every minor content update.
Off-Page SEO Terms
Off-page work concerns how other sources describe, reference and link to a website or business. Evaluate relevance and authenticity rather than treating every mention or link as equal.
Backlink. A link from another website to yours, also called an inbound link. Its context, source and destination help determine its usefulness.
Black Hat SEO. Industry shorthand for practices intended to manipulate rankings in ways that violate search-engine policies. The term is informal; the actual policies define the prohibited behaviour.
Broken Link Building. Finding relevant broken links on other sites and suggesting a useful replacement. The replacement should serve the original reference’s purpose.
Citation Building. Creating or correcting business references in relevant directories and platforms. In local SEO, a citation commonly includes business identity and contact information.
Cloaking. Showing different content to search engines and users with the intention of manipulating rankings or misleading users. Normal responsive layouts are not automatically cloaking.
Digital PR. Earning relevant coverage through research, stories, expertise or useful resources. Coverage can produce links and awareness, but neither is guaranteed.
Disavow File. A file asking Google to disregard specified links. Google’s disavow guidance reserves it for limited circumstances, rather than routine cleanup of every unfamiliar backlink.
Domain Authority (DA). Moz’s proprietary prediction score for a domain’s ranking potential. It is a third-party metric, not a Google score or a direct measure of trustworthiness.
Domain Rating (DR). Ahrefs’ metric describing a domain’s backlink-profile strength relative to its database. DR is not a Google ranking factor or a complete assessment of a site’s quality.
Editorial Link. A link selected by a publisher because it supports the surrounding content. Assess the actual context rather than assuming every editorial link has equal value.
Guest Post. An article contributed to another publication. Paying for links or using large-scale guest posting to manipulate rankings can breach search spam policies.
Link Building. Earning or acquiring references that help people discover useful content. Relevant resources and relationships are part of link-building work.
Link Equity. Informal shorthand for ranking value or signals associated with links. It is not a directly observable amount transferred according to a public formula.
Link Exchange. An arrangement to link between sites. Excessive exchanges intended to manipulate rankings differ from ordinary relevant references between organisations.
Link Spam. Links created primarily to manipulate search rankings. Google’s spam policies include paid ranking links and excessive link exchanges.
NAP. Name, address and phone number. Accurate business details help customers and platforms identify the business, but NAP consistency is not a complete local ranking strategy.
Page Authority (PA). Moz’s proprietary prediction score for an individual page’s ranking potential. It is distinct from Google’s internal evaluation of that page.
Private Blog Network (PBN). A network of websites used to manufacture links for ranking manipulation. Control of multiple websites alone does not define a PBN.
Referring Domain. A unique website domain linking to a page or site. One referring domain can supply many backlinks.
Sponsored Link. A link associated with payment or another commercial arrangement. Google recommends rel=”sponsored” to identify these relationships.
UGC Link. A link qualified with rel=”ugc” to indicate user-generated content, such as a comment or forum post.
Unlinked Mention. A reference to a business or resource without a hyperlink. It can support awareness, but should not be reported as a backlink.
White Hat SEO. Informal shorthand for SEO practices intended to follow search-engine policies and serve users. It is not an official certification.
SEO Measurement & Experience Terms
Measure search visibility alongside what visitors do after arriving. Our SEO measurement guide explains why one traffic or visibility metric cannot describe the whole journey.
Attribution. Assigning credit for an outcome to observed interactions. An attribution model cannot account perfectly for untracked visits, private sharing or every earlier influence.
Average Engagement Time. In GA4, time the website was in focus or the app was in the foreground, averaged using the report’s specified denominator. It is not necessarily total elapsed visit duration.
Average Order Value (AOV). Revenue divided by orders within a defined period. Use a consistent definition for refunds, tax and delivery charges.
Average Position. In Search Console, an average based on the topmost result from the measured property or page for an impression. Query mix and result layouts affect interpretation.
Bounce Rate. In GA4, the percentage of sessions that were not engaged sessions. It does not simply count people who view one page and immediately leave.
Click. A recorded selection that sends someone from a search result to an external page, subject to the platform’s rules. Search Console’s definitions vary across result types.
Click-Through Rate (CTR). Clicks divided by impressions, expressed as a percentage. Compare like-for-like queries, devices, positions and result types where possible.
Conversion. A business-relevant action, such as a completed purchase or qualified enquiry. Define the action explicitly so that minor interactions are not mistaken for outcomes.
Conversion Rate. Conversions divided by a defined population, such as sessions or users. Reports using different denominators are not directly comparable.
Cost Per Click (CPC). Advertising cost divided by paid clicks. A keyword tool’s CPC estimate describes an advertising market, not the value of every organic visit.
Customer Lifetime Value (CLV). An estimate of the value a customer contributes over a relationship. State whether it refers to revenue, gross profit or another measure.
Engaged Session. In GA4, a session that exceeds the configured engagement-time threshold, records a qualifying key event, or has at least two page or screen views. The default time threshold is 10 seconds.
Engagement Rate. Engaged sessions divided by sessions in GA4. GA4 defines engagement and bounce together, so the two percentages sum to 100%.
Event. A recorded interaction or occurrence in analytics, such as a page view or form submission. The event’s name alone does not prove correct implementation.
Google Analytics 4 (GA4). Google’s event-based analytics product for websites and apps. Its session, engagement and outcome definitions differ from Universal Analytics.
Generative AI Performance Report. A Search Console report showing impressions in supported Google generative search experiences. Its documented metrics do not provide separate clicks, CTR, positions or prompts; its impressions are already included in the Web performance report.
Impression. A counted appearance under a platform’s reporting rules. It is not necessarily a unique person or proof that someone read the result.
Key Event. A GA4 event marked as especially important to the business. Marking an event as key does not make it a qualified lead or sale.
KPI. Key performance indicator, a measure chosen to evaluate a business objective. A useful KPI has a definition, owner and decision it supports.
Organic Traffic. Visits attributed to unpaid search under the analytics system’s channel rules. Analytics sessions and Search Console clicks measure different things and need not match exactly.
Page View. A recorded display of a page. Repeat views and automatic event behaviour can affect totals.
Qualified Lead. An enquiry that meets agreed criteria such as service fit, location and genuine buying interest. A contact-page visit alone is not a qualified lead.
Referral Traffic. Visits attributed to links from other websites under the analytics platform’s rules. Search, social and other channels may be classified separately.
Return on Investment (ROI). Net return relative to investment. Document which revenue, margins and costs are included before comparing campaigns or channels.
Scroll Depth. How far down a page an interaction reaches, as recorded by a tracking setup. It can indicate exposure to content without proving comprehension.
Session. A group of interactions treated as one visit under an analytics system’s rules. Timeout settings and tracking implementation affect session counts.
Source/Medium. Analytics fields describing where traffic came from and the type of channel, such as a search engine and organic search.
UTM Parameters. URL parameters used to label campaigns for analytics. Use a consistent naming system and avoid applying campaign tags to ordinary internal links.
Year-on-Year (YoY). Comparing a period with the equivalent period one year earlier. Align dates and account for holidays, tracking changes and unusual events.
Zero-Click Search. A search that does not produce a click to an external website within the measurement used. It can end in an answer, a refinement or activity within the search platform.