AI marketing has its own vocabulary, and MCP adds another layer. This glossary defines the terms you'll encounter when working with AI marketing tools, reading about them, or setting up MCP connections.
Terms are alphabetical. Where something is commonly confused with something else, we've noted it. Where a term connects to Ooty's tools, we've linked to the relevant page.
Numbers You'll See Throughout This Glossary
The most-cited AI marketing statistics -- each linked to its original source in the definitions above
content production speed increase with AI writing tools
Loopex Digital, 2025
Source: Multiple sources, 2025 | ooty.io
A
A/B Testing. Comparing two versions of a page, ad, or email to see which performs better. Split your audience, show each group a different version, measure a defined outcome.
Agentic AI. An AI system that takes sequences of actions autonomously to complete a goal, rather than generating a single response per prompt. An agentic AI might research a topic, draft content, check it against your style guide, and send it for review without being prompted at each step.
AI Overview (Google). The AI-generated summary appearing at the top of some Google search results, above the traditional links. Powered by Gemini. Your brand only appears here if Google's systems select it as a credible source, which is distinct from ranking on page one.
AI Search. Search interfaces powered by LLMs that generate answers instead of returning ranked links. ChatGPT search, Perplexity, Google AI Mode, Copilot search. Also called generative search.
AI Visibility. How often a brand appears in AI-generated answers across ChatGPT, Perplexity, Google AI Overviews, and Gemini. A nascent metric that is closely tied to search ranking, since most LLMs search Google to generate their answers (see QFO). Measurement tools exist but are immature as of early 2026. Check yours with the free AI domain trust scorer.
Algorithm. A set of rules a system uses to make decisions. In marketing: the rules deciding who sees what content (social algorithms) or which pages rank for which queries (search algorithms).
Anchor Text. The clickable text in a hyperlink. Signals to search engines what the linked page is about.
Answer Engine. A search system returning direct answers rather than links. Perplexity explicitly describes itself this way. Traditional search engines are adding answer engine features (Google AI Overviews, Bing Answers).
Most marketing teams use AI the same way: copy data from one dashboard, paste it into ChatGPT, Gemini, or Claude, ask a question, and hope the answer is useful. The AI does its best with a screenshot or a CSV export, but it is working with a snapshot, not your
You keep seeing "MCP" in AI threads and newsletters. Everyone talks about it like you already know what it means. Here is the short version, then we will set it up.
MCP in 30 seconds
MCP (Model Context Protocol) is an open standard that lets AI assistants conn
Model Context Protocol (MCP) connects AI assistants like ChatGPT, Claude, and Gemini to your live marketing data. Instead of copying numbers from Google Analytics into a chat window, your AI reads the source directly. Anthropic released MCP in November 2024. P
API (Application Programming Interface). A defined way for one system to communicate with another. MCP servers typically wrap existing APIs to make them accessible to AI assistants.
Attribution. Crediting the touchpoints that led to a conversion. Common models: last-click, first-click, linear, data-driven.
Average Position. The mean rank at which your page appears in Google results for a query. An average of 1.0 means you always appear first. Used in Ooty SEO to find ranking opportunities.
B
Backlink. A link from one website to another. Links from credible sites signal authority to search engines. Also relevant for AI visibility: backlinks help you rank in Google, and LLMs primarily cite pages that rank well in the search results they retrieve.
BERT. A language model Google uses in its search algorithm. One of the first major shifts toward semantic understanding in search. A precursor to current LLMs.
Bid Strategy. In paid ads, the approach for determining how much you pay per click, impression, or conversion. AI-powered bid strategies (Google Smart Bidding, Meta Advantage+) adjust in real time.
Bounce Rate. The percentage of sessions where a user visits one page and leaves. In GA4, the equivalent is "engagement rate" (sessions with 10+ seconds, multiple pages, or a conversion).
Broad Match. A Google Ads keyword match type where your ad shows for searches related to your keyword, including synonyms and paraphrases.
C
Cache. A temporary data store. Some MCP servers cache API responses to reduce latency and avoid rate limits. Cached data may be slightly out of date.
ChatGPT. OpenAI's AI assistant. Relevant to marketing because it drives the majority of AI referral traffic, brands want to be recommended when users ask for suggestions, and it supports MCP via its Agents SDK.
Citation (AI Search). When an AI system references a source by linking to it in a generated response. AI systems typically cite 2 to 7 domains per response, far fewer than a Google results page.
Click-Through Rate (CTR). Clicks divided by impressions, times 100. Low CTR despite high impressions in Search Console usually means your title or description isn't compelling enough.
Claude. Anthropic's AI assistant. Natively supports MCP connections. Most MCP servers, including marketing-focused ones, work across Claude, ChatGPT, Gemini, and other compatible clients.
Consent Mode (Google). A framework for adjusting how Google's tracking tags behave based on cookie consent choices. Essential for GA4 compliance under GDPR and similar regulations.
Content Score. A numerical measure of how well content is optimised for a target keyword, from tools like Surfer SEO or Clearscope. Based on comparison with top-ranking pages.
Conversion. A desired action: a purchase, form submission, sign-up, download. Conversion rate is conversions divided by sessions, times 100.
Core Web Vitals. Google's page experience metrics: Largest Contentful Paint (loading), Interaction to Next Paint (interactivity), Cumulative Layout Shift (visual stability). A ranking factor. Accessible through Ooty SEO.
Cost Per Acquisition (CPA). Total spend divided by conversions.
Cost Per Click (CPC). Amount paid per ad click. Tracked in Ooty Ads.
Cost Per Mille (CPM). Cost per 1,000 impressions. Standard for brand awareness campaigns.
Crawl Budget. The number of URLs Googlebot will crawl on your site in a given timeframe. Large sites may have important pages left uncrawled if budget is wasted on low-value URLs.
CRM. Software tracking customer interactions and sales pipelines. HubSpot has an official MCP server; Salesforce connects via Composio.
D
Data Freshness. How recently data was fetched. Critical in MCP contexts because some servers cache API responses. Good implementations note the timestamp.
Domain Authority / Domain Rating. Third-party metrics (DA from Moz, DR from Ahrefs) estimating a domain's authority based on backlinks. Not Google metrics, but useful for relative comparison.
E
E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness). Google's framework for evaluating content quality. Experience means first-hand knowledge; Expertise means subject matter depth; Authoritativeness means reputation; Trustworthiness means accuracy. Matters for both SEO and GEO.
Embedding. A numerical representation of text that captures semantic meaning. AI systems use embeddings to find relevant content for a query. The backbone of RAG systems.
Engagement Rate. The percentage of users who interact meaningfully with content. Platform definitions vary significantly.
Entity. A real-world thing (person, place, organisation, product) that AI systems recognise and can reason about. Making sure AI systems have accurate information about your brand entity improves GEO performance.
F
Feature Gate. A mechanism to enable or disable functionality for specific users. MCP servers use feature gates to control which tools are visible based on licence type or plan.
Fine-Tuning. Training a pre-trained model on a specific dataset for a particular task. Distinct from prompting: fine-tuning modifies the model's weights.
First-Party Data. Data collected directly from your audience. Increasingly important as third-party cookies disappear.
G
GA4 (Google Analytics 4). Google's current analytics platform. Event-based rather than session-based. Accessible via Ooty Analytics.
Generative AI. AI systems that create new content: text, images, audio, video, code. GPT-4, Gemini, Claude, and Stable Diffusion are all generative AI.
Generative Engine Optimisation (GEO). Optimising content to appear in AI-generated answers from ChatGPT, Perplexity, and Google AI Overviews. Best understood as a format layer on top of SEO rather than a separate discipline, since LLMs primarily cite content they find via traditional search engines (see QFO). See our GEO vs SEO guide.
Gemini. Google's family of AI models. Powers Google AI Overviews, Workspace AI features, and Google's AI search products.
Google Indexing API. Lets site owners notify Google when a page is added or removed, prompting faster crawling. Available through Ooty SEO.
How MCP Fits Together
The three pieces of the MCP architecture and how they connect
MCP Client
Your AI assistant
Claude, ChatGPT, Gemini
MCP Server
The connector
Ooty SEO, Ahrefs, HubSpot
Tools & Data
Your actual data
Search Console, GA4, YouTube
From your perspective: You ask Claude a question. Claude calls the MCP server. The server fetches your data. You get a real answer. The architecture is invisible.
Source: Model Context Protocol, modelcontextprotocol.io | ooty.io
H
Hallucination. When an AI generates false information presented as fact. MCP tools reduce hallucination risk for data tasks by pulling from real sources rather than generating from training data.
I
Impression. A single instance of content being displayed. In search: your page appeared in results. In ads: your ad was served.
Impression Share. In Google Ads, the percentage of impressions received vs. total eligible. Low share due to rank means bids are too low; due to budget means you're running out before the day ends.
Intent (Search Intent). What a user really wants from a search. Four types: informational, navigational, commercial, transactional. Matching content to intent is fundamental to both SEO and GEO.
J
JSON-LD. A format for embedding structured data in web pages, used for schema markup. Google's recommended format.
JSON-RPC. The communication protocol underlying MCP. When your AI calls an MCP tool, it sends a JSON-RPC request to the server. You don't need to understand this to use MCP.
K
Keepa. A service tracking Amazon product pricing and rank history. Used by Ooty Commerce.
Keyword. A word or phrase users type into a search engine. In SEO, you target keywords with content. In paid search, you bid on them.
Keyword Difficulty. A 0 to 100 score estimating how hard it would be to rank on page one. Available in Ahrefs, Semrush, and similar tools. Not a Google metric.
Knowledge Graph (Google). Google's database of entities and relationships. Getting into it improves brand visibility in both traditional search and AI answers. Accessible via Ooty SEO.
L
Landing Page. The page a user arrives on after clicking a link. In paid ads, a dedicated page designed to convert visitors from a specific campaign.
Latency. Time between making a request and getting a response. Most MCP servers respond in 1 to 5 seconds.
LCP (Largest Contentful Paint). A Core Web Vital. Under 2.5 seconds is good, over 4 seconds is poor.
LLM (Large Language Model). An AI model trained on large amounts of text to understand and generate language. GPT-4, Gemini, Claude, and Llama are all LLMs.
Long-tail Keyword. A keyword with three or more words. Lower volume, higher intent, lower competition. Typically converts better than head terms.
M
MCP (Model Context Protocol). An open standard that lets AI assistants connect to external data sources and tools. MCP servers provide data; MCP clients (ChatGPT, Gemini, Claude) consume it. See our MCP guide for marketers.
MCP Client. An AI application that connects to MCP servers. ChatGPT, Gemini, Claude, Cursor, and others.
MCP Server. A connector exposing data or tools to MCP clients. Ooty builds MCP servers for marketing data sources.
MCP Tool. A specific action an MCP server offers. Examples: get_top_queries, get_page_impressions, compare_periods.
Meta Description. The snippet below a page title in search results. Not a ranking factor, but affects CTR. Check yours with the free meta analyzer.
Multi-Touch Attribution. Giving credit to multiple touchpoints in the customer journey rather than just the first or last.
N
Natural Language Processing (NLP). AI focused on understanding and generating human language. Powers chatbots, sentiment analysis, content analysis.
Natural Language Query. A question phrased in ordinary language. MCP enables natural language queries against marketing data: instead of building a custom report, ask your AI "which landing pages have the worst bounce rate?"
O
OAuth. An authorisation framework letting users grant an application access to their data without sharing passwords. When you connect Ooty to Google, you're completing an OAuth flow.
Organic Search. Unpaid search results. Data available in Search Console and GA4; AI-powered analysis via Ooty SEO.
P
PageRank. Google's original algorithm for ranking pages based on inbound link quality and quantity. Still a core component, though vastly more complex now.
PageSpeed Insights. Google's tool for measuring page performance against Core Web Vitals. Data accessible via Ooty SEO.
Perplexity. An AI search engine that generates answers with cited sources.
Prompt. An instruction or question given to an AI system.
Prompt Engineering. Crafting inputs to get better, more reliable outputs. For marketing: be specific about audience and goal, provide context, specify format.
Q
Quality Score (Google Ads). A 1 to 10 rating based on expected CTR, ad relevance, and landing page experience. Higher scores mean lower CPCs and better positions. See our complete Quality Score guide for how to check, improve, and diagnose each component.
Query. A search term entered by a user. Different from a keyword: keywords are what you target, queries are what users typed.
Query Fan-Out (QFO). The process by which LLMs generate search queries and send them to traditional search engines (usually Google) to retrieve results before synthesizing an answer. When you ask ChatGPT a question, it doesn't search its own index. It searches Google, gets back up to 8 results, and works with those. This is why ranking in Google is the primary driver of AI visibility. See our GEO vs SEO guide for the full explanation.
R
RAG (Retrieval-Augmented Generation). An AI architecture where the model retrieves relevant information from an external source before generating a response. MCP is one approach for enabling retrieval.
ROAS (Return on Ad Spend). Revenue divided by ad spend. A ROAS of 3 means you earned 3 for every 1 spent.
Rate Limit. A restriction on API requests per time period. MCP servers must handle them gracefully. Large data pulls may be slower because the server paginates.
Remote MCP. An MCP server hosted remotely (via URL) rather than running on your computer. No local installation needed, just a connection URL.
S
Schema Markup. Structured data vocabulary annotating web pages so search engines and AI systems can understand them precisely. Article, Product, FAQ, HowTo are common types. Validate yours with the free schema validator.
Search Console. Google's tool showing how your site appears in Google Search. The most accurate source of organic search data for your own site. Core to Ooty SEO.
Semantic SEO. Creating content that covers a topic thoroughly rather than repeating a keyword. Google's language understanding makes semantic relevance more important than keyword frequency.
Sentiment Analysis. AI analysis of whether text is positive, negative, or neutral. Used for reviews, comments, surveys, brand mentions.
Smart Bidding. Google Ads' AI-powered bid strategies that adjust automatically toward target CPA, ROAS, or conversion goals.
Structured Data. Data in a defined, machine-parseable format. AI systems prefer structured data over prose when extracting facts.
T
Temperature. A parameter controlling AI output randomness. Low (0 to 0.3): predictable, factual. High (0.7 to 1.0): creative, varied. Use low for data analysis, high for creative work.
Token. The basic unit of text processing in LLMs. Roughly 0.75 words. AI APIs typically charge per token.
Tool (MCP). A specific capability an MCP server exposes. Your AI discovers available tools when it connects, then calls them as needed during conversation.
Topical Authority. A ranking signal based on whether your site demonstrates deep expertise across a topic, not just individual pages on individual keywords.
U
UA (Universal Analytics). The previous Google Analytics, replaced by GA4 in July 2023.
V
Value Proposition. A clear statement of what benefit a product delivers and why someone should choose it over alternatives.
W
Webhook. A mechanism for one application to notify another when an event occurs. In marketing automation, webhooks trigger actions in connected systems.
Y
YouTube Analytics. Performance data for YouTube channels. Views, watch time, retention, subscribers, thumbnail CTR. Accessible through Ooty Video.
Z
Zero-Click Search. A search where the user gets their answer from the results page without clicking through. Featured snippets, knowledge panels, and AI Overviews all drive this. One reason GEO matters alongside traditional ranking.
Zero-Shot Prompting. Asking an AI to perform a task without examples. For complex marketing tasks, few-shot prompting (providing 2 to 5 examples) typically works better.
Where These Terms Connect to Real Tools
Each Ooty product touches multiple concepts defined in this glossary
Ooty SEOSEO & Search
Related terms: Average Position, CTR, Impressions, PageSpeed, Knowledge Graph, Search Console
