AI92 is live - our global soft launch is on: the full platform is open worldwide for the first 30 days.

ResourcesGlossary

Glossary

The AI marketing glossary

67 plain-English definitions across AI search, SEO, reviews, outreach, campaigns and APIs - no hype, and no jargon left unexplained.

AI search & answer engines

AI Visibility
How often - and how prominently - a brand appears in AI-generated answers from assistants such as ChatGPT, Gemini or Perplexity. It is the AI-era counterpart of rankings: instead of tracking positions on a results page, you track whether models mention or cite you when people ask relevant questions. See also: SEO & AI Visibility
AEO (Answer Engine Optimization)
The practice of structuring content so answer engines - systems that reply with a single synthesized answer instead of a list of links - can find, understand and reuse it. Typical AEO work includes direct question-and-answer formatting, structured data and concise, quotable summaries.
GEO (Generative Engine Optimization)
Optimizing content to be retrieved and cited by generative AI systems such as ChatGPT, Perplexity or Google's AI Overviews. It overlaps heavily with AEO; the term stresses influencing what generative models say about a topic or brand, not just where a page ranks.
Answer engine
A search system that answers questions directly instead of returning ranked links - Perplexity, ChatGPT with browsing, or Google's AI Overviews. Answer engines synthesize replies from sources they retrieve, which is why being retrievable and citable now matters as much as ranking.
LLM (Large Language Model)
A machine-learning model trained on very large amounts of text to understand and generate language - the technology behind ChatGPT, Claude and Gemini. LLMs power answer engines, chat assistants and most modern AI marketing tools.
LLM citation
When an AI assistant names or links a specific source in its answer. Citations are how brands earn visible credit in AI search: the cited page gets the mention, and - in engines that link out - the click. See also: SEO & AI Visibility
AI Overviews
Google's AI-generated summary block shown above the classic results for many queries. It synthesizes an answer from several sources and links a handful of them, which changes how much attention and traffic the standard listings below receive.
RAG (Retrieval-Augmented Generation)
A technique where an AI model first retrieves relevant documents and then writes its answer grounded in them. Answer engines are essentially RAG systems running over the web - which is why crawlable, well-structured pages are what get quoted.
Hallucination
When an AI model states something false with confidence - an invented fact, source or number. Hallucinations are the core risk of unreviewed AI content; grounding answers in retrieved sources and keeping a human approval step are the standard mitigations.
llms.txt
A plain-text file served at a site's root that gives AI systems a concise, machine-readable overview of what the site is and where its key content lives. It complements robots.txt: robots.txt says what may be crawled, llms.txt says what is worth reading. See also: Developer hub
AI crawler
A bot that fetches web content for AI systems - for model training (like GPTBot) or for answering live queries (like PerplexityBot). Sites declare access policies for these crawlers in robots.txt, separately from classic search-engine bots.

SEO fundamentals

SEO (Search Engine Optimization)
The discipline of improving a site's visibility in organic - unpaid - search results, spanning technical health, content relevance and authority signals such as links. Modern SEO increasingly includes visibility in AI-generated answers, too. See also: SEO & AI Visibility
SERP (Search Engine Results Page)
Everything a search engine shows for a query: classic links, ads, maps, AI summaries, images and more. The non-link blocks are called SERP features; ranking work is about winning the right spots on the right SERPs.
Keyword
A word or phrase people type - or say - when they search, used as the basic unit of demand in SEO. Keyword research maps which queries matter for a business, how often they are searched and how hard they are to win.
Long-tail keyword
A longer, more specific query with lower search volume but clearer intent and weaker competition - 'emergency plumber open sunday' versus 'plumber'. Long-tail terms usually convert better because the searcher already knows exactly what they want.
Search intent
The goal behind a query: informational (learn something), navigational (reach a specific site), commercial (compare options) or transactional (buy now). Matching a page's format and content to intent is one of the strongest levers in search.
Organic traffic
Visits that arrive from unpaid search results, as opposed to ads, social or direct visits. It is the compounding return on SEO work: a ranking keeps sending visitors without a per-click cost.
Domain authority
A third-party estimate - Moz's DA, Ahrefs' DR and similar - of how strong a site's link profile is, usually on a 0–100 scale. Useful for comparing sites at a glance, but it is not a metric Google itself uses.
Crawling
How search-engine and AI bots discover pages: following links, reading sitemaps and fetching content. If a page cannot be crawled it cannot be indexed or cited, which makes crawlability step zero of any visibility work.
Indexing
A search engine storing and organizing a crawled page so it can appear in results. Crawled does not mean indexed: engines skip pages they consider thin, duplicated or blocked - a noindex tag, for instance, keeps a page out on purpose.
Canonical URL
The 'official' URL a site declares for a piece of content when several URLs serve the same or near-identical page. The rel=canonical tag consolidates ranking signals onto one URL and prevents duplicate-content dilution.
hreflang
An annotation that tells search engines which language and region each version of a page targets, so searchers get the right locale in their results. Multilingual sites - this one serves 52 locales - use hreflang to stop language versions competing with each other.
Structured data (schema.org)
Machine-readable annotations, usually JSON-LD, that label what is on a page: a product, a review, an FAQ, a local business. Search engines use it to power rich results, and answer engines use it to extract facts reliably.
Meta description
The short page summary in HTML that search engines often display under the title in results. It is not a ranking factor, but it strongly influences click-through - it is effectively the page's ad copy on the SERP.
Core Web Vitals
Google's user-experience metrics for pages: LCP (loading speed), INP (responsiveness to interaction) and CLS (visual stability). They are measured on real visits and feed Google's page-experience signals.
Local SEO
Optimizing to appear for searches with local intent - 'near me' and city-level queries - centred on the Google Business Profile, consistent local citations, reviews and locally relevant content. See also: Manage Google Reviews
Local pack
The map block with a few business listings that Google shows for local-intent queries. Winning a spot depends on relevance, distance and prominence - with review count, rating and profile completeness playing a large role.

Reviews & Google Business Profile

Google Business Profile (GBP)
Google's free business listing - formerly Google My Business - that controls how a business appears on Maps and in local search: hours, photos, posts, Q&A and reviews. It is the foundation of local visibility. See also: Manage Google Reviews
GBP post
A short update published directly on a Google Business Profile - an offer, an event, a piece of news - that shows on the listing itself. Regular posting keeps a profile visibly active and gives searchers a current reason to choose you. See also: Manage Google Reviews
Review velocity
The rate at which a business earns new reviews over time. A steady stream signals an active, trusted business to both consumers and ranking systems, while a sudden unexplained spike can look manipulated. See also: Reviews & Reputation
Review response rate
The share of customer reviews a business replies to. Responses - especially to negative reviews - are read by every future customer browsing the profile, and they signal that someone is actually listening.
Star rating
The average of a business's review scores, usually on a one-to-five scale. Small differences matter disproportionately around common filter thresholds, and recent ratings typically weigh more with readers than old ones.
Reputation management
The ongoing practice of monitoring what is said about a brand across reviews, social media and search - and acting on it: responding, resolving issues and fixing root causes. Legitimate reputation work never fabricates sentiment. See also: Reviews & Reputation
Review gating
Screening customers by sentiment first and only asking the happy ones for a public review. It violates Google's review policies - and most other platforms' rules - and can get reviews filtered or a listing penalized. Ask everyone, uniformly, instead.
NAP consistency
Keeping a business's Name, Address and Phone number identical everywhere it appears - website, Google Business Profile, directories. Inconsistent NAP data fragments local ranking signals and erodes trust in the listing.

Outreach & email deliverability

Cold outreach
Contacting prospects who have no prior relationship with you, usually by email. Done well it is targeted, personalized, low-volume and easy to opt out of; done badly it is spam - and mailbox providers treat it accordingly. See also: Outreach
ICP (Ideal Customer Profile)
A precise description of the company or person who gets the most value from your product: industry, size, role, geography, buying triggers. Prospect lists, messaging and qualification are all built against the ICP.
Contact discovery
Finding the right people to reach at target companies - names, roles and contact details - from public and licensed data sources. Quality beats volume: invalid addresses damage sender reputation with every send. See also: Contact Finder
Lead enrichment
Adding attributes to a contact or company record - size, industry, role, location - so you can segment, personalize and route leads properly. Enrichment turns a bare email address into a record you can act on. See also: Contact Finder
Email deliverability
The likelihood that your email lands in the inbox rather than the spam folder - or nowhere at all. It is driven by sender reputation, authentication (SPF, DKIM, DMARC), list hygiene, sending patterns and recipient engagement. See also: Outreach
SPF (Sender Policy Framework)
A DNS record listing which servers are allowed to send email on behalf of your domain. Receiving servers check it to catch forged senders; it is one of the three standard email-authentication mechanisms, alongside DKIM and DMARC.
DKIM (DomainKeys Identified Mail)
A cryptographic signature added to outgoing email that lets receiving servers verify the message really comes from your domain and was not altered in transit. A practical prerequisite for deliverability at any meaningful volume.
DMARC
A DNS policy that tells receiving servers what to do with mail that fails SPF or DKIM alignment - deliver, quarantine or reject - and where to send reports. Major mailbox providers require it for bulk senders.
Email warm-up
Gradually increasing sending volume from a new domain or mailbox so providers learn it is a legitimate sender. Jumping straight to high volume from a cold domain is one of the fastest routes to the spam folder.
Bounce rate (email)
The share of sent emails that could not be delivered: hard bounces (the address does not exist) and soft bounces (temporary failures). A high bounce rate signals a stale or scraped list and quickly damages sender reputation.
Open rate
The share of delivered emails that were opened. Useful as a trend, but privacy features - Apple Mail's pre-fetching, for example - inflate it, so treat open rate as directional rather than exact.
Spam trap
An email address that exists only to catch senders with bad list practices: it never opted in to anything, so any mail it receives proves the list was not cleanly built. Hitting spam traps can get a sender blocklisted.
Unsubscribe rate
The share of recipients who opt out after a send. A small, steady rate is normal list hygiene; spikes flag mistargeted content - and hiding the unsubscribe link is both illegal in most jurisdictions and disastrous for deliverability.

Campaigns, creative & analytics

CTA (Call to action)
The specific next step a piece of marketing asks for: book a demo, start a trial, get directions. One clear CTA per asset almost always outperforms several competing ones.
A/B testing
Showing two variants of an asset - subject line, page, creative - to comparable audiences and measuring which performs better. Valid tests change one variable at a time and run long enough to reach statistical significance.
Conversion rate
The share of visitors or recipients who complete the desired action: sign up, book, buy. Define the denominator per channel and funnel stage, or the numbers stop being comparable.
CTR (Click-through rate)
Clicks divided by impressions - or, for email, clicks divided by delivered messages. CTR measures how compelling the message was to the audience that actually saw it; it is the first honest engagement signal in most funnels.
UTM parameters
Tags appended to URLs - utm_source, utm_medium, utm_campaign and friends - so analytics tools can attribute a visit to the campaign that caused it. Consistent UTM conventions are the difference between attribution and guesswork.
Brand kit
The reusable definition of a brand's identity: logo, colors, fonts, tone of voice. It keeps every designed or AI-generated asset consistent - creative tools use the brand kit as a constraint when producing content. See also: Creative Studio
Content calendar
A schedule of what will be published, where and when, across channels. It turns publishing from ad-hoc bursts into a steady cadence - which is what both algorithms and audiences reward. See also: Campaign Architect
Social proof
Evidence that other people already trust you: reviews, ratings, testimonials, customer logos, usage numbers. It works because buyers outsource risk assessment to peers - and it only works if it is real, current and verifiable.

APIs & credits

API (Application Programming Interface)
A programmatic interface that lets software call another system's capabilities directly, without a user interface. AI92 exposes its marketing modules through a REST API documented in the developer hub. See also: Developer hub
API key
A secret token that identifies and authenticates whoever is calling an API. Treat keys like passwords: keep them server-side, never ship them in client code or repositories, and rotate any key that may have been exposed. See also: Authentication docs
Bearer token
An HTTP authentication scheme where the caller sends 'Authorization: Bearer <token>' with each request. Bearer means possession is proof - anyone holding the token can use it - so tokens must only ever travel over HTTPS. See also: Authentication docs
Sandbox key
An API key scoped to a test environment: requests behave like production but touch no real data, budgets or third-party services. Build and test integrations on sandbox keys, then switch to live keys for launch. See also: Sandbox docs
Credit metering
A usage-billing model where each API action debits a defined number of credits from a balance. Metering makes the cost of every action explicit up front and allows spend to be capped per key, feature or period. See also: Credits & metering docs
Pay-as-you-go
Paying for actual usage instead of a fixed subscription - for example, buying credit packs that API calls consume. It suits spiky or exploratory workloads; spend caps and auto-recharge rules keep it predictable. See also: API pricing
Rate limiting
A cap on how many requests a client may make in a given time window, protecting the platform and enforcing fair use. Well-behaved clients read the rate-limit response headers and slow down before hitting a 429 error. See also: Rate limits docs
Webhook
A callback the platform sends to your server when something happens - a job finishes, a contact is verified - so you do not have to poll for updates. Endpoints should verify the signature, respond quickly and tolerate the occasional duplicate delivery. See also: Webhooks docs
Idempotency
The property that repeating the same request has the same effect as making it once. APIs implement it with idempotency keys, letting clients retry safely after a timeout without double-charging or duplicating work. See also: Idempotency docs
OpenAPI specification
A machine-readable description of a REST API: endpoints, parameters, schemas, authentication. It powers generated documentation, client SDKs and tests - AI92's public API reference and pricing tables are generated from its live spec. See also: API reference