Glossary
Glossary of search, frontend, and AI terms
41 terms defined in plain language, grouped by discipline. Where a term maps to real work, the entry links to the page that covers it properly.
9 terms
SEO
- Canonical tag
- An HTML link element that tells search engines which URL is the authoritative version among duplicate or near-duplicate pages. It consolidates ranking signals to one URL, which matters for faceted navigation, tracking parameters, and syndicated content.Technical SEO
- Citation consistency
- The degree to which a business name, address, and phone number match exactly across directories, review sites, and data aggregators. Inconsistent citations weaken local ranking signals and can confuse the algorithms that reconcile duplicate business records.Local SEO tactics
- Core Web Vitals
- A set of Google metrics measuring real-world page experience: Largest Contentful Paint for loading speed, Interaction to Next Paint for responsiveness, and Cumulative Layout Shift for visual stability. They are a ranking input and a proxy for whether visitors can actually use the page.Technical SEO
- Crawl budget
- The number of pages a search engine is willing and able to crawl on a site within a given period, driven by server capacity and perceived site value. Sites with large catalogs or heavy duplicate content can waste crawl budget on low-value URLs, delaying discovery of new or updated pages.Technical SEO
- Indexation
- The process by which a search engine stores a crawled page in its index so it can be returned in results. A page can be crawled but excluded from the index due to noindex tags, thin content, or duplication signals, which is why crawling and ranking are separate problems.Technical SEO
- Internal linking
- The practice of linking between pages on the same site to distribute authority, establish topical relationships, and help both users and crawlers reach important content. A deliberate internal linking structure is one of the few ranking levers a site fully controls.On-page optimization
- Local pack
- The map-based block of three business listings Google shows above organic results for location-relevant queries. Ranking in it depends on Google Business Profile completeness, proximity, and review and citation signals rather than the standard organic ranking factors.Google Business Profile
- Pillar and cluster
- A content architecture where one broad pillar page covers a topic at a high level and links out to narrower cluster pages that each target a specific subtopic, with clusters linking back to the pillar. It concentrates topical authority and gives crawlers a clear map of related content.Content strategy
9 terms
AEO
- Answer engine optimization
- The practice of structuring content so AI assistants and chat-based search tools can extract and cite it directly as an answer, rather than only ranking it in a list of links. It emphasizes clear entity definitions, structured data, and self-contained passages over traditional keyword targeting.AI search visibility
- Entity
- A distinct, uniquely identifiable thing such as a person, organization, product, or place that search engines and language models track independently of the words used to describe it. Entity clarity, established through consistent naming and structured data, helps AI systems disambiguate a business from similarly named competitors.Entity optimization
- Extractability
- How easily a specific fact or claim can be pulled out of a page as a self-contained, unambiguous statement by an AI system. Content buried in long paragraphs with unclear antecedents is harder to extract than a direct sentence stating a fact with its subject named explicitly.Answer engine audit
- Grounding
- The technique of anchoring an AI-generated answer to specific retrieved sources so the response can be traced back to verifiable content instead of the model's unconstrained output. Well-grounded content is written in extractable, unambiguous passages that a model can quote or paraphrase safely.AI citation tracking
- Knowledge graph
- A structured database of entities and the relationships between them, used by search engines to answer questions directly and to disambiguate similarly named entities. Appearing correctly in a knowledge graph depends on consistent structured data and citations across authoritative sources.Entity optimization
- llms.txt
- A proposed plain-text file, placed at a site's root, that gives AI crawlers a concise, structured summary of a site's key pages and purpose. It is analogous to robots.txt but aimed at helping language models understand and prioritize content rather than controlling crawl access.llms.txt implementation
- Retrieval augmented generation
- An architecture where a language model's response is grounded by first retrieving relevant documents or passages from an external index, then generating an answer conditioned on that retrieved text. It reduces fabrication by tying output to verifiable source material rather than relying solely on model memory.AI search visibility
- Structured data
- Markup, typically in JSON-LD using schema.org vocabulary, that explicitly labels page elements such as organizations, products, reviews, or FAQs for machines. It removes ambiguity that a crawler or model would otherwise have to infer from prose, improving eligibility for rich results and AI citations.Schema markup
7 terms
Application frontends
- Design tokens
- Named, reusable values, such as a specific color, spacing unit, or font size, stored once and referenced throughout a design system and codebase instead of being hardcoded repeatedly. Changing a token updates every component that references it, keeping a large application visually consistent as it grows.Design systems
- Edge runtime
- A constrained JavaScript execution environment that runs application code on servers geographically close to the requesting user, rather than in one central data center. It lowers latency for server-rendered pages and API responses but limits access to some Node.js APIs.Internal tools
- Hydration
- The process of attaching interactive JavaScript behavior to server-rendered HTML once it reaches the browser, turning static markup into a fully functional application. Poorly optimized hydration is a common cause of pages that look loaded but are unresponsive to clicks for a noticeable delay.SaaS frontends
- Route loader
- A function tied to a specific application route that fetches the data that route needs before or during rendering, rather than fetching it inside a component after the page has already mounted. It reduces loading-spinner cascades and keeps data-fetching logic colocated with the route it serves.Dashboards
- Row level security
- A database-enforced access control model where permission rules are evaluated per row, so a query automatically returns only the records a given user or tenant is allowed to see. It is central to building multi-tenant CRMs and client portals without duplicating authorization logic in every application query.Client portals
- Server side rendering
- A rendering approach where HTML is generated on the server for each request and sent to the browser already populated, rather than being built client-side after a blank page loads. It improves initial load performance and gives crawlers and AI agents complete content without requiring JavaScript execution.Migration from static
- State management
- The set of patterns and tools used to store, update, and synchronize application data across components, distinct from local UI state that lives in a single component. Choosing the right approach, whether server state caching or a global client store, is a major factor in how maintainable a custom dashboard or CRM becomes.Custom CRM
6 terms
AI and automation
- Agent
- A system that uses a language model to decide which actions to take, in what order, toward a goal, often calling external tools or APIs and adapting based on their results. This distinguishes an agent from a simple chatbot, which only produces text in response to a prompt.Autonomous agents
- Fine tuning versus context
- Two different ways to adapt a language model's behavior: fine tuning retrains the model's weights on custom examples, while providing context supplies relevant information directly in the prompt at run time. Context-based approaches are usually cheaper and faster to update, which is why most business automations favor them over fine tuning.AI training
- Human in the loop
- A workflow design where a person reviews or approves an AI system's output at a defined checkpoint before it takes effect, rather than letting the system act fully autonomously. It is standard practice for automations touching customer communication, financial data, or published content.Automation workflows
- Large language model
- A neural network trained on large volumes of text to predict and generate language, underlying tools like chat assistants, drafting tools, and coding copilots. Its output is a statistical continuation of patterns in training data, which is why it requires grounding and human review for factual or business-critical tasks.Frontier AI services
- Prompt injection
- An attack where malicious instructions are hidden inside content a language model processes, such as a webpage or document, in an attempt to override its original instructions. It is a real risk for any agent that reads untrusted external content and is a reason to constrain what actions an agent can take autonomously.Agentic workflows
- Workflow orchestration
- The coordination of multiple tools, data sources, and steps, including AI and non-AI systems, into a single reliable process with defined triggers, error handling, and sequencing. It is what turns a one-off script or prompt into a repeatable business process that runs without manual intervention.Automation tools
5 terms
Paid media
- Attribution model
- The rule set used to assign credit for a conversion across the multiple touchpoints, such as ads and organic visits, that preceded it, for example last-click, first-click, or data-driven attribution. The model chosen materially changes which channels look effective and can shift budget decisions significantly.Conversion tracking
- Negative keyword
- A term or phrase explicitly excluded from triggering a given ad, used to prevent spend on searches that look relevant but do not convert, such as job seekers searching a service the business sells. Ongoing negative keyword maintenance is one of the most reliable ways to cut wasted spend in a search campaign.Google Ads
- Performance Max
- A Google Ads campaign type that automatically places ads across Search, Display, YouTube, Gmail, and Maps from a single campaign, with the algorithm deciding budget allocation across those channels. It trades granular manual control for automated optimization, which requires strong conversion tracking to steer it toward the right outcomes.Google Ads
- Quality score
- Google Ads' 1 to 10 estimate of how relevant a keyword, ad, and landing page are to a searcher, which directly affects both ad rank and cost per click. A low quality score means paying more for the same ad position, so it is a direct lever for lowering acquisition cost.Google Ads
- Retargeting frequency
- The number of times an individual user is shown a retargeting ad within a given period, a setting used to balance staying visible against ad fatigue and brand annoyance. Frequency caps are usually set per platform and adjusted based on how far along the funnel the audience segment sits.Retargeting
5 terms
Analytics
- Cohort analysis
- A method of grouping users by a shared starting point, such as the week they first signed up, and tracking how their behavior diverges over time. It reveals retention and lifetime value trends that a single blended metric across all users would hide.Analytics insights
- Deduplication
- The process of ensuring a single conversion or event is not counted more than once across overlapping tracking systems, such as both server-side and client-side tags firing for the same purchase. Unresolved duplication inflates reported conversions and undermines trust in the data feeding paid media decisions.Conversion tracking
- Event taxonomy
- A documented, consistent naming and parameter structure for every tracked event across a site or app, agreed on before implementation. Without one, teams end up with duplicate or inconsistently named events that make cross-channel reporting unreliable.GA4 setup
- Key event
- Google Analytics 4's term for a conversion, an event marked as representing meaningful business value such as a form submission or purchase. Marking the wrong events as key events, or leaving important ones unmarked, is a common cause of misleading conversion reporting.Conversion tracking
- Statistical significance
- A measure of how likely an observed difference between two results, such as two ad variants or landing pages, is due to a real effect rather than random chance. Acting on a test before it reaches significance is one of the most common ways teams draw the wrong conclusion from an experiment.Strategy and optimization
What is the difference between SEO and AEO?
SEO earns a ranked position on a results page. AEO earns a citation inside a generated answer. They share foundations like clean rendering and clear entities, but AEO also depends on whether a passage can be extracted and attributed on its own.
- A page can rank well and never be cited, and the reverse also happens.
- Extractability is a structural property, not a keyword one.
- Both fail immediately if the server does not render the content.
Related
Where these terms get used
Want someone who speaks this fluently on your side
Start with a diagnosis. Plain language, written findings, no jargon for its own sake.