The AI search glossary

A side-by-side comparison of two versions running at the same time. Decisions come from data instead of guesswork.

Content that everyone can use, including people with disabilities. Pages built for accessibility also read better to machines.

Writing and structuring content so it can be lifted straight into an answer. Question headings, clean definitions, and short paragraphs are the core of it.

An AI system that plans its own steps and uses tools once you give it a goal. It can run searches, fill out forms, and complete purchases.

The schema type that carries a product's average rating and review count.

An application that talks with people in natural language, answers questions, and can carry out tasks. Where search, recommendation, and comparison behavior now lives.

Automated crawlers AI companies use to collect content or fetch pages at answer time. GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and CCBot are common examples.

How cleanly a machine can read and understand a page. Server-rendered text, a clear heading hierarchy and structured data all raise it.

Visits that arrive from links inside AI interfaces. In analytics they show up under referrers like chatgpt.com or perplexity.ai.

A measurable read on how often a brand appears in AI assistant answers, in what context, and in what tone.

Text that describes what an image shows. It serves accessibility and tells machines what is in the picture.

The visible text a link sits on. It tells both people and engines what the target page is about.

The share of your tracked prompt set where the brand makes it into the answer. The breadth dimension of visibility.

A search interface that resolves a query into a single answer. Perplexity and Google AI Overviews are the common examples.

The brand's slice of all mentions across AI answers. Share of voice, translated to generative search.

The same prompt returning different answers on different runs. It comes from the probabilistic nature of language models.

The interface two systems use to exchange data. Content, product, and measurement data all flow automatically through this layer.

The review steps content clears before it goes live. At enterprise brands that includes legal and brand sign-off.

The schema type for articles and news content. It carries author, publish date, and updated date fields.

A touchpoint that contributes to a conversion without being the last click. AI answers usually play this role.

The share of catalog products with a given field filled in. The core measure of feed quality.

The model that decides which channel gets credit for a conversion. First click, last click, linear, and data-driven are the common ones.

The information showing who wrote a piece and what qualifies them. Both a trust signal and a structured data field.

Whether a product can be bought right now. Without real-time updates, ad budget flows to products that are out of stock.

Average revenue per order. One of the core inputs to category and campaign decisions.

Where the brand lands, on average, when it appears in a list. Beyond making the list, it shows how much weight the brand carries on it.

A link to your site from another site. In traditional SEO it signals authority; in GEO it raises the odds of being picked as a source.

The first measurement taken before any work starts. Every later gain is read against it.

The reference value you compare against. It can be the category average, the category leader, or your own past performance.

Systematic tendencies a model carries over from its training data. It can skew how brands and categories are represented.

Microsoft's search engine. Because it feeds Copilot and, for a period, ChatGPT web results, its effect on visibility runs above its market share.

Deciding which automated traffic your security layer lets through. Rules that are too tight can block AI bots too.

How well your audience knows the brand. Unaided awareness is recalling the name unprompted; aided awareness is recognizing it from a list.

A model inventing facts about your brand. Wrong prices, products you do not sell and incorrect company details all count.

The overall state of a brand across awareness, perception, and preference.

The measurable rise in awareness, preference, and purchase intent after a campaign or program. Usually measured by survey.

The brand named in the text of an answer. Even with no link attached, a mention is the first sign you made the consideration set.

The coherent story a brand tells about itself. It is the source AI draws on when it summarizes you.

A question that names the brand directly. Used to see how the brand is described and whether anything in the description is wrong.

Keeping the brand out of contexts it does not belong in. In AI answers this shows up as unflattering comparisons and old crises resurfacing.

Measuring awareness, preference, and perception at regular intervals. Traditionally by survey, now with digital signals as well.

The consistent tone, word choice, and sentence structure in a brand's writing. Leave it undefined and content at scale stops sounding like one brand.

How many times a month people search the brand name. For brands that appear in AI answers without getting clicks, this is the first number to move.

The schema type that shows where a page sits in the site hierarchy. It spells out your category structure for machines.

A link pointing to an address that does not exist. Users and bots both hit a wall.

Producing hundreds of pages in one pass. The most common approach for category copy and product descriptions.

Content that tells the reader what to weigh when deciding. It answers the comparison need without naming brands.

Temporary storage for frequently requested content. Set up wrong, it can serve bots stale content.

Two or more of your pages competing for the same topic. The signals split, so none of them lands strongly.

The tag that declares the primary address when the same content sits at several URLs.

The share of users who add to cart and leave without buying. Shipping, payment, and lack of trust are the main causes.

Content that reports a real customer result in numbers. One of the content types AI engines read as a trust signal.

The overall score a feed earns for completeness, accuracy, and freshness.

The descriptive copy on an ecommerce category page. It introduces the category to shoppers and defines its scope for AI.

A question that describes the category without naming a brand. What are the best running shoes belongs in this group.

A network that serves content from servers geographically close to the user. It cuts latency.

A model working through the steps to an answer out in the open. It improves accuracy on complex reasoning.

The point or percentage difference between two measurement periods. Reports usually show two: change against the previous period and against the baseline.

OpenAI's chat interface. With web search on, it adds source links to its answers; with it off, it answers from what the model already knows.

Splitting long content into meaningful pieces a model can work with. Sections set off by a heading and complete on their own chunk better.

The share of customers who leave in a given period. It is the defining metric of a subscription business.

How ready a piece of content is to be cited by an AI. Clear definitions, dates, hard numbers, and a named author all raise it.

The share of answers mentioning your brand where your own site is cited as the source. This is the metric that turns a mention into traffic.

An AI answer crediting a source by name or by link. Citations generate both visibility and referral traffic.

Anthropic's language model family and assistant interface. Widely used for long-context and document analysis work.

Clicks divided by impressions. Source links in generative answers get clicked less, so a falling rate alone does not mean failure.

Building the page in the browser with JavaScript. Flexible for user experience, risky for machine readability.

The cycle where measurement finds the gap, content closes it, and the new content gets measured again. It is what turns GEO work into more than a one-off report.

A measure of how much content shifts while a page loads. Ads and images without declared dimensions are the main causes.

A group of users who share a start date or a defining trait. Used to see how behavior changes over time.

Searches close to purchase that involve comparison and price. The query group nearest to conversion.

Content that weighs options criterion by criterion. Pages with a criteria table surface more often on comparison queries.

A question that weighs two or more options. AI usually answers these with a list or a table.

Systematically gathering information on how competitors position, publish, and show up.

How visibility is split among competitors on the same prompt set. Your own score only means something against this spread.

The set of options a buyer holds in mind while deciding. AI answers now shape that set directly.

The production spec for a piece of content: its goal, its audience, the questions it covers, its length, and its tone.

The plan for what publishes when. It lines up with campaign and seasonal cycles.

Merging pages that cover the same topic in fragments into one strong page.

Topics your audience asks about that your site has no answer for. In GEO terms, the gap surfaces in the questions where AI cannot describe you.

Comparing your current content inventory against the target topic list and prioritizing what is missing.

A table of every piece of content on the site with topic, type, date, and performance. Every content decision starts here.

Creating demand with content that gives value instead of pitching a sale.

Removing or merging pages that create no value. It raises the average quality of the site.

Updating the data, examples, and date on an existing page. It usually returns more than producing something new.

Growing content volume while holding the same quality bar. Without templates, briefs, and an approval workflow, scaling just costs you quality.

The total number of tokens a model can hold in view at once. Anything that does not fit in the window cannot reach the answer.

A visitor completing the action you are after. A purchase, a form fill, or a demo request are the common ones.

Conversions divided by sessions or users. This is the ratio that shows what AI-driven traffic is actually worth.

Improving pages and flows to get more conversions from the traffic you already have. Testing your way forward is the point.

A broad update to a search engine's ranking systems. It sends wide swings through the results.

Google's core set of page experience metrics. It covers loading, interactivity, and visual stability.

Marketing that manages corporate reputation and perception rather than product sales.

How many pages a bot will crawl on your site in a given window. On large ecommerce sites it is a hard constraint.

Bots following links to collect pages. AI bots use the same mechanism.

The total marketing and sales spend it takes to win one customer.

The sequence of every step where a customer meets the brand. AI assistants have moved into the research stage of that journey.

The total net revenue a customer generates over the life of the relationship.

The rating and comment a customer leaves on a product. It shapes how AI answers describe that product.

A document that defines every field, unit, and value in use. Keeps teams aligned on feed and content projects.

The gap between an event happening and it showing up in a report. In AI measurement it can run a few days, depending on measurement frequency.

A model family known for its open weights. The cost advantage makes it a common pick for in-house deployments.

The schema type that marks up a term and its definition. On glossary pages, it makes each term machine-readable.

Building interest and need for a category. Demand capture, by contrast, serves need that already exists.

A prospect asking to see the product live. In a sales-led model, it is the main conversion step in the funnel.

Earning media coverage with content that carries real news value. Because most AI answers draw on third-party sources, it has a direct GEO effect.

A site's address on the internet. AI answers usually credit sources at the domain level.

A third-party score that estimates the overall strength of a domain. It is not an official Google metric.

The same text living at more than one address or across more than one site. Ecommerce sites that paste manufacturer descriptions hit this most often.

A slot in a template that fills automatically from product or category data. Attributes like brand, size, and material map into these fields.

Showing shoppers ads for the exact products they viewed. It runs on catalog and pixel data.

Google's experience, expertise, authoritativeness, and trust framework, written for its quality raters. Author info, cited sources, and company transparency all strengthen it.

Media placement you get without paying for it. News coverage, reviews, and mentions all count.

How accurate, original, and readable content is, and how well it holds the brand voice. At scale, it gets scored and tracked.

A representation that turns the meaning of text into a numeric vector. Texts with similar meaning sit close together in vector space.

The single, machine-recognized identity behind a brand, person, place, or concept. Modern search runs on entities, not words.

Telling apart different entities that share a name. Critical in markets crowded with similar brand names.

Pulling generated content out of the system. Necessary for teams that publish through their own stack.

The sprawl of URLs created by filter and sort options. Left unchecked, it spawns thousands of pointless pages.

A section listing real customer questions with clear answers. Keeping question and answer side by side lifts citability directly.

The schema type that marks up question and answer pairs. It makes content easier to reuse as an answer.

The short answer box above search results. It is the forerunner of generative summaries and rewards the same content structure.

A single data field inside the feed. Title, brand, color, size, and material are common examples.

A data problem that gets a product disapproved by the platform. Missing identifiers, price mismatches, and policy violations are the main causes.

Transformation logic that turns raw data into the format a platform expects. Adding the brand to a title or mapping categories happens here.

Putting a few examples in the prompt so the model locks onto the format you want. With no examples at all, it is called zero-shot.

Training an existing model on extra data for a specific task or domain.

How often a brand is named first in an answer. The first option carries a clear advantage in user behavior.

Google's language model family and assistant interface. Tightly wired into data across the Google ecosystem.

A system that answers a question with text instead of a list of links. It can draw on both model knowledge and live web data.

The practice of getting a brand to appear, be described accurately, and be cited as a source in the answers generative AI engines like ChatGPT, Perplexity, and Gemini produce.

A systematic read of where a brand stands across AI engines. Visibility, citation sources, factual errors, and content gaps are reported together.

Google's fully conversational search mode. It goes deeper through follow-up questions and leads with an answer instead of a list of links.

The generative summary block above Google search results. It pushes classic organic results down and changes click behavior.

A business's official listing on Google. Address, hours, and reviews all feed from here.

The dashboard where you upload product data to Google and get it approved. Shopping ads and free listings run from here.

Where a product sits in Google's standard category taxonomy. A wrong match distorts both competition and impressions.

Google's free performance and indexing tool for site owners. It is the primary source for impression, click, and crawl issue data.

xAI's assistant. Its access to real-time posts on X can produce different results on current events and social topics.

Tying an answer to a verifiable source. The main method for cutting hallucination.

A product's globally unique barcode number. It lets platforms identify and compare the product.

The rule layer that blocks outputs a model should not produce. Used for brand safety and compliance requirements.

A model stating something untrue with full confidence. It happens more often where information is thin or contradictory.

A heading structure that starts at H1 and steps down in logical order. It makes it easier for a model to split content into sections.

Content written for people first that carries real information. Pages built only for search engines lose ground under this standard.

Content that walks through a task step by step. AI tends to reproduce it as a numbered list.

The schema type that marks up step-by-step instructions.

A tag that links the language and country versions of the same content. On multi-market sites it makes sure the right version is served.

The code a server returns with its response. 200 means success, 301 a permanent redirect, 404 not found, 410 permanently removed.

The encrypted connection protocol. Security is table stakes, so its absence costs trust immediately.

An editor checking AI-generated content before it goes live.

Turning the static HTML the server sent into an interactive page in the browser. Set up correctly, it keeps both speed and readability.

A company-level definition of the customer type that produces the most value. Drawn with criteria like industry, size, and maturity.

Preparing images for search through file name, alt text, size, and format.

How many times a piece of content or a link is shown to a user. In AI interfaces, most engines do not expose impression data at all.

The extra result that would not have happened without the work. The most honest measure of real contribution.

Adding a crawled page to the search engine's database. A page that is not indexed appears for no query at all.

Running a trained model to produce an answer. Cost and latency both land at this stage.

How long the interface takes to respond after a user interaction. It measures that frozen feeling after a click.

Connecting the pages of a site to each other. It is the cheapest way to show topic relationships to users and engines alike.

Text and links that code injects after the page loads. Because they are absent from the source, many bots never read them.

A way to add structured data to a page as a separate script. It is the recommended method because it leaves the visual design untouched.

A small, agreed-upon set of metrics that shows how close you are to the goal.

The phrase someone types into a search box. On the GEO side, the equivalent is the prompt.

A score for how hard it is to rank near the top for a keyword. It is a tool's estimate, not an exact figure.

The date a model's training data ends. Anything after it reaches the model only through web access.

A structure that stores entities and the relationships between them. It determines what search engines understand a brand to be.

A model trained on enormous volumes of text that writes by predicting the next word. It is the engine behind every generative assistant.

The time between request and response. Long context and larger models push it up.

How long the largest content element on the page takes to appear. It is the moment a user counts the page as loaded.

Optimizing content so language models understand and remember it correctly. Another name for the work GEO describes.

A proposed file at the site root that gives language models a plain-text summary of the site and its most important pages.

A question with a location in it. On prompts like best private hospital in Ankara, AI leans harder on local sources and map data.

Building visibility in location-based searches. On local questions, AI weights map and business listing data heavily.

The schema type for businesses with a physical location. It carries address, hours, and service area.

Adapting content to the language and culture of a target market. Unlike literal translation, units, examples, and search habits change too.

Reading server logs to see which bot visits which page and how often. It is the most reliable way to observe AI bot behavior.

Narrow searches that each draw little volume but run to enormous numbers. Most AI questions are long tail by nature.

A statistical method for separating out how much each channel contributes to sales. It runs without cookies, which is why it gained ground as tracking got harder.

A platform where multiple sellers list products. Their product pages are often more visible than the brand's own site.

The open protocol that connects AI applications to external data sources and tools in a standard way.

How often the prompt set is run again. Weekly is a balanced starting point for most brands.

The percentage of tracked prompts where the brand name appears. The most basic measure of visibility.

The hierarchy of a primary message and the messages that support it. It keeps every channel consistent.

Meta's assistant, built into WhatsApp, Instagram, and Facebook. Huge reach, limited source attribution.

The product catalog behind Facebook and Instagram ads. It is the data source for dynamic ads.

The short summary of a page shown in search results. Not a ranking signal, but it moves click-through rate.

The page's title tag. It is the first line shown in search results and states the page's topic in one line.

The older method that places structured data inside the HTML tags themselves. Harder to maintain than JSON-LD.

Microsoft's assistant, built on Bing. Because it ships inside Windows and Office, it reaches enterprise users at scale.

Tracing wrong brand information in AI answers back to its source and replacing it with accurate, accessible content.

A page that works cleanly on small screens. Evaluation runs largely on the mobile version.

Reporting results from the same prompt set split out by model. A brand that is strong in ChatGPT and weak in Gemini is common.

A document laying out a model's capabilities, limits, and intended uses.

A standard set of tests built to measure the quality of a model's output. The GEO equivalent is the prompt set.

The company that builds a model and ships it through an API or an interface. OpenAI, Google, Anthropic, xAI, and Meta are the main providers.

Releases of the same model family with different dates and different capabilities. A version change is the most common cause of unexplained jumps in brand visibility.

Revenue that repeats every month in a subscription model. The annual equivalent is ARR.

The part number a manufacturer assigns to a product. It serves as the identifier for products without a GTIN.

A model that handles images, audio, and video alongside text. It brings product images and videos into scope for analysis.

A one-question metric for how likely customers are to recommend the brand.

One number: the percentage of positive mentions minus the percentage of negative ones. Negative values call for immediate action.

An attribute that declares a link passes no authority. The sponsored and ugc values cover ad and user-content links.

A directive that keeps a page out of search results. Applied in the wrong place, it makes pages disappear quietly.

Movement that comes from fluctuation in measurement conditions rather than real change. It is the most common source of false readings in small prompt sets.

The schema type that defines a product's selling terms. Price, currency, and availability live here.

A model whose weights are published for download. Because you can run it in-house, it fits cases where data privacy matters.

Visits that arrive from search results without ads. AI summaries are pushing it down on informational queries.

The schema type that describes a company. Brand name, logo, contact details, and official accounts are declared here.

Content that offers information or a point of view other sources do not. Pages that only repeat what is already out there never get picked as a source.

The channels a brand controls. Site, blog, email list, and social accounts all count.

How fast a page loads. Slow pages hurt both the user and crawl efficiency.

Splitting long lists across multiple pages. Built wrong, some products never get crawled.

Visibility bought with budget. Ads and sponsorships fall here.

A model for working with agencies and consultancies. Reseller, referral, and co-delivery arrangements are the common forms.

The related-questions block in search results. A practical source for the questions real users ask.

Google's campaign type that runs automatically across all inventory. Creative and feed quality decide the outcome.

An answer engine that shows a source list with every answer. Its source transparency makes it the reference interface for GEO measurement.

A fictional profile standing in for the target customer. Defined by needs, objections, and decision criteria.

A question where the user describes themselves first. As a new mom, which product should I buy belongs in this group.

The hub page that covers a topic end to end and points to the supporting pieces. It lets AI learn the topic from a single source.

Tracking code that reports on-site user behavior back to an ad platform.

Where a brand stands in its category: which need it serves, for whom, and against what.

The price in the feed matching the price on the site exactly. One of the most common reasons for disapproval.

The schema type that describes a product. Covers name, brand, identifier, price, and availability.

Copy covering a product's features, use, and what sets it apart. AI comparison answers feed directly off this text.

Filling in and improving thin or missing product data. Descriptions, attributes, and images all count.

A file or stream that carries product title, price, availability, and image data in a structured format. Ad platforms and marketplaces feed off it.

A structured data set that passes a product's price, availability, and shipping details to the search engine.

Structuring the title so brand, product type, and distinguishing attributes appear in a set order.

The seller's own category path. Used for reporting and campaign segmentation.

The text a user types to an AI assistant. In GEO measurement, the prompt plays the role the keyword plays in classic SEO.

Structuring a prompt to get the output you want. Assigning a role, specifying a format, and giving examples are the most common techniques.

The list of prompts a brand tracks on a schedule, chosen to represent its category and its customers' questions.

Running the same prompts against different models at set intervals, then recording and comparing the answers.

Managing a brand's relationship with the press and the public. On the digital side it turns directly into source visibility.

The defined set of steps content moves through from brief to publish. It answers who writes, who approves, and where it ships.

The stages from awareness to purchase. Each stage calls for different content and a different kind of query.

A prospect that meets the agreed criteria and is judged ready for sales.

What the user is trying to do with a search. Four main groups: informational, navigational, commercial investigation, and transactional.

A model retrieves the relevant documents before it answers, then grounds the answer in them. This is what makes answers current and sourced.

Where a page sits for a given query. In generative answers, ranking gives way to the order of mention in a list.

Capping how many requests are accepted in a given window. Past the cap, requests are rejected.

Another standard for embedding semantic markup in HTML. Adoption is limited.

How easily text is understood. Short sentences, clear headings, and plain words make reading easier for people and models alike.

The list of brands or products an AI returns in answer to a question. Making the list and ranking high on it are two different goals.

Sending one URL to another. Use 301 for permanent moves and 302 for temporary ones.

A URL that reaches its final destination through more than one hop. Every hop adds latency and loses signal.

The number of unique sites linking to you. A more meaningful signal than total link count.

Getting the same result under the same conditions. Full reproducibility is hard with language models, so measurements are repeated.

One page that rearranges itself to fit the screen size. It removes the need for a separate mobile site.

The step that finds the document chunks closest to the question. Content that cannot be retrieved cannot reach the answer.

The schema type for a single review. Includes author, rating, and body fields.

The expanded search listing that structured data makes possible. Ratings, prices, and question lists are common examples.

Reinforcement learning from human feedback: people score model outputs and the model is tuned toward those scores. This is the stage that makes assistants helpful and consistent.

The revenue returned for every unit of ad spend.

A file at the site root that tells each bot which directories it may crawl. You can set separate rules for AI bots.

A model where the purchase runs through a sales conversation. In product-led growth, the user starts on their own.

The schema field that points to a brand's other official profiles. Social accounts, Wikipedia, and Wikidata links go here.

The number of prompts and repeats behind a measurement. A small sample makes noise look like a real swing.

The shared structured data vocabulary search engines use. It turns what is on the page into tags machines can read.

The average number of monthly searches for a keyword. Useful for setting priorities, but not a measure of value on its own.

A subset of data split by a shared attribute. Category, device, source, and persona splits are the common ones.

Writing markup with tags that carry meaning. Heading, list, table, and article tags hand the structure straight to machines.

Search that matches on meaning rather than words. It is the basis for how generative engines work.

A numeric measure of how close two texts are in meaning. Different words can still score high.

Whether answers describe the brand in positive, neutral, or negative terms. The quality dimension of visibility.

The work of raising a site's organic visibility in search engines. It stands on three legs: content, technical foundations, and authority.

A search engine results page. Today organic results share it with generative summaries, ads, and rich results.

HTML is assembled on the server and sent to the browser ready to render. Bots see the content without running JavaScript.

One uninterrupted visit by a user to the site. The base unit of analytics reporting.

Your share of all branded searches in the category. Used as an early indicator of market share.

The percentage of total category visibility a brand holds. In traditional marketing it is calculated on ad spend, in GEO on mentions in answers.

An ad format that shows the product image and price. It is built straight from feed data.

How pages sit relative to each other and how they link. Good architecture gets both users and bots to the right page in fewer steps.

A machine-readable file listing the site's important URLs and their update dates. It keeps bots from skipping pages.

Any signal showing what other people experienced. Reviews, ratings, testimonials, and case studies all count.

The distribution of which domains an AI answer draws on for a topic. Your own site, competitors, media, and forums each read separately.

A difference large enough that chance does not explain it. In small prompt sets, most week-to-week movement is not significant.

Running the schema markup you added through testing tools to confirm it is valid and error-free.

The instruction layer that sets how a model behaves, invisible to the user. Part of the answer difference between interfaces starts here.

The setting that controls randomness in model output. Higher values add creativity, lower values add consistency.

A predefined structure for content of the same type. Heading pattern, section order, and length are fixed by the template.

The share of sources cited in answers that sit on sites the brand does not own.

The smallest unit a model works with when processing text. A single word often spans several tokens.

A model getting work done by calling external systems. Running a search, executing code, or hitting an API all count.

A group of interlinked content built around one main topic. A pillar page sits at the center with subtopics around it.

A hierarchical list of every topic a brand wants to own. It decides up front which page covers which topic.

The trust a brand earns by publishing thorough, consistent content on one subject. It drives which sources AI answers pull from.

Every point where a customer runs into the brand. Ads, search results, stores, and now AI answers.

The pool of text a model learns from. Getting your content into that pool means long-term visibility.

Rewriting the wording of a source text while keeping its meaning intact. Preferred over translation for campaign and brand copy.

The written version of audio or video. Models work on text, so a transcript is what makes the content visible.

The architecture underneath language models. Its attention mechanism catches relationships across long text.

The direction a metric moves over time. A single measurement is a snapshot, the trend is the real signal.

The time from sending a request to receiving the first byte. An indicator of server and infrastructure performance.

A brand name that appears without a link back to the source. It is the most common kind of mention in AI answers.

Building URLs so they read cleanly, stay consistent, and reflect the site hierarchy.

Content your customers create: reviews, questions, photos. AI answers lean on it as a trust signal.

The one-sentence answer to why a customer should pick you.

The color, size, and other options of the same product. Set variant handling up wrong and you get both inventory and content problems.

A database that stores embedding vectors and searches them by semantic similarity. The retrieval layer of RAG architectures.

Making video findable through titles, descriptions, transcripts, and structured data.

The schema type that describes video content. Carries duration, thumbnail, and transcript data.

A single composite number that rolls up mentions, citations, and ranking. Used to compare across periods and against competitors.

A model's ability to consult the live web while it answers. On, you get current information and source links; off, it works from training data alone.

A mechanism that pushes an automatic notification to another system the moment an event fires. Real-time flow without constant polling.

An open, machine-readable knowledge base. Many models and search systems draw entity data from it.

Sending generated content straight to WordPress as a draft or a live post. Removes the copy-paste step.

Topics that directly affect a person's life, such as health, finance, and law. The quality bar in these areas is markedly higher.

The user gets the answer inside the search interface and ends the search without clicking through to any site. AI answers have made this markedly more common.