Glossary

AI marketing glossary

56 terms from paid media, analytics, SEO, AI search and automation, defined in plain language. Every entry links to the post that goes deeper.

56 terms

A

A/B test CRO & Websites

An A/B test shows two versions of a page, ad or email to comparable audiences at the same time and measures which one produces more of a defined outcome. It only produces a trustworthy answer when there is enough traffic for the difference to be more than noise, so low-traffic sites are usually better off applying a clear best-practice fix directly.

Read: CRO basics: what actually moves revenue →

Advantage+ Paid Media

Advantage+ is Meta's family of automated campaign and creative features, where Meta's systems choose audiences, placements, budgets and sometimes creative variations for you. Its creative enhancements can alter what people actually see, so they should be reviewed setting by setting rather than left on by default.

Read: A Meta Ads creative testing framework you can actually maintain →

AI agent AI & Automation

An AI agent is a system where a language model decides which steps to take and which tools to call to reach a goal, rather than following a fixed sequence. Reliable agents keep hard business rules in deterministic logic and give the model only the decisions that genuinely need judgement.

Read: Building your first AI agent: what it is and what it isn't →

AI Mode SEO & AI Search

AI Mode is Google Search's conversational, AI-generated search experience, where answers are composed by a Gemini model and users can ask follow-up questions. It now carries ads in a growing list of countries, which makes visibility inside generated answers a paid media question as well as an SEO one.

Read: AI SEO vs traditional SEO: what actually changes →

AI Overviews SEO & AI Search

AI Overviews are the AI-generated summaries Google shows above traditional results for some searches, with links to the sources used. Being cited in an Overview depends on the same fundamentals as ranking, plus content that states answers clearly enough to be lifted.

Read: How to get your business cited by ChatGPT and Perplexity →

AI SEO SEO & AI Search

AI SEO is the work of making a website easy for AI assistants such as ChatGPT, Perplexity, Gemini and Google's AI Overviews to find, understand and cite. It builds on traditional SEO rather than replacing it, adding entity clarity, quotable statements and machine-readable structure.

Read: What AI SEO actually means, and the four things I changed on my own site →

API AI & Automation

An API (application programming interface) is a defined way for one piece of software to request data or actions from another. In marketing automation, APIs are how a workflow reads leads from a form tool, writes them to a CRM or sends a prompt to an AI model.

Read: Webhooks explained for marketers (no code required) →

Attribution Analytics

Attribution is the method used to decide which marketing touchpoints get credit for a conversion. Every ad platform attributes generously to itself, so the sum of platform-reported conversions is almost always higher than the real number of sales.

Read: Attribution: why every platform over-reports, and what to do about it →

Attribution window Analytics

An attribution window is the period after a click or view during which a platform will claim a conversion, for example 7 days after a click. Longer windows report more conversions without any change in actual sales, so windows must match before platforms are compared.

Read: Attribution: why every platform over-reports, and what to do about it →

B

Break-even ROAS Paid Media

Break-even ROAS is the return on ad spend at which advertising pays for itself after the cost of goods, calculated as 1 divided by the gross margin. A product with a 40% margin needs a ROAS of 2.5 just to break even, before any other costs.

Read: How to set a CPA target you can actually afford →

Broad match Paid Media

Broad match is the Google Ads keyword match type that lets an ad show for searches related to the keyword's meaning, not just its wording. It works best paired with Smart Bidding and accurate conversion tracking, and wastes money quickly without them.

Read: Google Ads campaign structure that actually scales →

C

Canonical tag SEO & AI Search

A canonical tag tells search engines which URL is the main version of a page when similar or duplicate versions exist. It consolidates ranking signals onto one URL instead of splitting them across copies.

Read: Schema markup that actually matters (and what to skip) →

Context window AI & Automation

A context window is the maximum amount of text, measured in tokens, that a language model can consider in a single request, including the prompt, attached data and its own reply. Sending more than the task needs raises cost and can lower answer quality.

Read: AI API costs explained: how to estimate before you build →

Conversion rate CRO & Websites

Conversion rate is the percentage of visitors or clicks that complete a defined action, such as a purchase or enquiry. Improving it lowers the cost per acquisition of every traffic channel at once, which is why CRO often outperforms extra ad spend.

Read: CRO basics: what actually moves revenue →

Conversions API Analytics

The Conversions API (CAPI) is Meta's server-to-server method for sending conversion events directly from a website's server or CRM, instead of relying only on the browser pixel. It recovers events lost to ad blockers and browser privacy restrictions and improves the signal Meta optimises on.

Read: Server-side tracking: do you actually need it? →

Core Web Vitals CRO & Websites

Core Web Vitals are Google's three user-experience metrics for page loading (Largest Contentful Paint), responsiveness (Interaction to Next Paint) and visual stability (Cumulative Layout Shift). They are a ranking signal and, more importantly, a direct driver of conversion rate on mobile.

Read: Core Web Vitals in plain language, and how to actually fix them →

CPA Paid Media

CPA (cost per acquisition) is total ad spend divided by the number of conversions, such as sales or qualified leads. A useful CPA target is worked back from margin and customer value, not copied from an industry average.

Read: How to set a CPA target you can actually afford →

CPC Paid Media

CPC (cost per click) is ad spend divided by clicks. It is an efficiency signal, not a success metric: a cheap click that never converts is more expensive than a costly one that does.

Read: The marketing metrics that actually matter (and the ones that don't) →

CPM Paid Media

CPM (cost per mille) is the cost of 1,000 ad impressions. On Meta and TikTok it mostly reflects auction competition and audience demand, and it rises sharply during peak retail periods such as Black Friday.

Read: The marketing metrics that actually matter (and the ones that don't) →

Creative fatigue Paid Media

Creative fatigue is the decline in an ad's performance as the same audience sees it repeatedly, visible as falling click-through rate and rising cost per result. The fix is a steady testing cadence of new concepts, not small tweaks to the same ad.

Read: A Meta Ads creative testing framework you can actually maintain →

CRO CRO & Websites

CRO (conversion rate optimisation) is the practice of increasing the share of visitors who take a desired action, through clearer messaging, less friction, stronger trust signals and faster pages. Structural fixes usually beat cosmetic redesigns.

Read: CRO basics: what actually moves revenue →

CTR Paid Media

CTR (click-through rate) is clicks divided by impressions. It indicates how relevant an ad or search listing is to the person seeing it, but it says nothing about what happens after the click.

Read: The marketing metrics that actually matter (and the ones that don't) →

E

E-E-A-T SEO & AI Search

E-E-A-T stands for experience, expertise, authoritativeness and trustworthiness, the qualities Google's quality rater guidelines use to assess content. It is not a single ranking factor, but first-hand experience and clear authorship are what separate useful content from generic content.

Read: AI content and Google: separating what's true from what's repeated →

Embeddings AI & Automation

Embeddings are numerical representations of text that place similar meanings close together, allowing software to search by meaning rather than by exact words. They power most retrieval systems that feed relevant documents to an AI model.

Read: Building your first AI agent: what it is and what it isn't →

Enhanced conversions Analytics

Enhanced conversions is a Google Ads feature that sends hashed first-party data, such as an email address from a checkout or lead form, alongside a conversion to improve matching. It recovers conversions that cookie-based tracking misses.

Read: Setting up GA4 conversion tracking properly →

Entity SEO & AI Search

An entity is a distinct, identifiable thing, such as a person, business, place or product, that search engines and AI models recognise independently of the words used to describe it. Consistent names, details and sameAs links across the web help them identify your business correctly.

Read: Schema markup that actually matters (and what to skip) →

F

Fine-tuning AI & Automation

Fine-tuning is further training of an existing AI model on your own examples so it adopts a particular format, style or task behaviour. For most marketing use cases, better prompts and retrieved context achieve the same result faster and cheaper.

Read: Prompt engineering for marketers: the parts that actually matter →

G

GA4 Analytics

GA4 (Google Analytics 4) is Google's event-based analytics platform, where every interaction is recorded as an event with parameters. Its numbers are only as good as the event and key event setup underneath them.

Read: Setting up GA4 conversion tracking properly →

Generative Engine Optimization SEO & AI Search

Generative Engine Optimization (GEO) is another name for AI SEO: optimising content to be retrieved, understood and cited by generative AI answer engines. The term is used interchangeably with AI SEO and answer engine optimisation (AEO).

Read: AI SEO vs traditional SEO: what actually changes →

H

Hallucination AI & Automation

A hallucination is a confident AI output that is false or unsupported, such as an invented statistic, citation or product feature. The practical defence is supplying source material, asking the model to say when it does not know, and checking anything that will be published.

Read: When not to use AI in your marketing →

I

Incrementality Analytics

Incrementality is the share of conversions that would not have happened without a given campaign. It is measured with holdout or lift tests and is the most honest answer to whether advertising is actually working.

Read: Attribution: why every platform over-reports, and what to do about it →

K

Key event Analytics

A key event is GA4's name for an event marked as important to the business, such as a purchase or a submitted lead form; it replaced the term conversion inside GA4. Only key events imported into Google Ads count as conversions for bidding.

Read: Setting up GA4 conversion tracking properly →

L

LLM AI & Automation

An LLM (large language model) is an AI model trained on very large amounts of text to predict and generate language, such as GPT, Claude and Gemini. LLMs are strong at drafting, summarising, classifying and extracting, and unreliable as a sole source of facts.

Read: When not to use AI in your marketing →

llms.txt SEO & AI Search

llms.txt is a plain markdown file placed at a website's root that gives AI systems a curated summary of the site and links to its most important pages. It is a proposed standard, cheap to implement, and makes a site easier for AI tools to navigate.

Read: llms.txt explained, and how to write one →

Local SEO SEO & AI Search

Local SEO is optimising a business to appear in location-based searches and Google Maps, driven mainly by a complete Google Business Profile, consistent business details across directories and a steady flow of genuine reviews.

Read: Local SEO for South African businesses: the parts that move the needle →

LTV Paid Media

LTV (customer lifetime value) is the total gross profit a customer generates over the whole relationship. Knowing it tells you how much you can afford to pay to acquire a customer, which is often far more than the first order suggests.

Read: How to set a CPA target you can actually afford →

M

MCP AI & Automation

MCP (Model Context Protocol) is an open standard for connecting AI assistants to external tools and data sources through a common interface. It lets one assistant read a CRM, a drive or an ad account without a custom integration for each.

Read: Building your first AI agent: what it is and what it isn't →

MER Paid Media

MER (marketing efficiency ratio) is total revenue divided by total marketing spend across every channel. Because it does not rely on any platform's attribution, it is the most reliable top-line check on whether marketing is paying back.

Read: The marketing metrics that actually matter (and the ones that don't) →

N

Negative keywords Paid Media

Negative keywords stop a Google Ads campaign from showing for searches you do not want, such as jobs, free, DIY or irrelevant locations. A pre-launch negative list prevents a large share of wasted spend in a new account.

Read: Google Ads campaign structure that actually scales →

P

Performance Max Paid Media

Performance Max is Google's automated campaign type that runs a single campaign across Search, Shopping, YouTube, Display, Gmail, Maps and AI Mode. It performs well with strong conversion data and good assets, and poorly as a substitute for either.

Read: Performance Max: when it works and when it quietly wastes money →

Prompt AI & Automation

A prompt is the instruction and context given to an AI model to produce an output. The most reliable prompts state the objective, supply the facts the model cannot infer, set constraints and define the output format.

Read: Prompt engineering for marketers: the parts that actually matter →

Prompt injection AI & Automation

Prompt injection is an attack where instructions hidden in content an AI processes, such as a web page, email or document, try to override the system's original instructions. Any automation that feeds outside content to a model with access to tools needs to treat that content as data, not commands.

Read: Building your first AI agent: what it is and what it isn't →

R

RAG AI & Automation

RAG (retrieval-augmented generation) is a method where relevant documents are retrieved first and passed to an AI model alongside the question, so the answer is grounded in your own sources. It is the standard way to make an AI answer accurately from company data.

Read: Building your first AI agent: what it is and what it isn't →

ROAS Paid Media

ROAS (return on ad spend) is revenue attributed to advertising divided by ad spend, so R4,000 of revenue from R1,000 of spend is a ROAS of 4. It ignores margin, which is why a high ROAS can still lose money.

Read: The marketing metrics that actually matter (and the ones that don't) →

S

sameAs SEO & AI Search

sameAs is a schema.org property that links an entity's structured data to its other official profiles, such as LinkedIn, Instagram or Wikipedia. It helps search engines and AI models confirm that those profiles all describe the same person or business.

Read: Schema markup that actually matters (and what to skip) →

Schema markup SEO & AI Search

Schema markup is structured data, usually in JSON-LD format, that describes a page's content to machines using the schema.org vocabulary, for example an article's author or a business's address. It removes ambiguity for search engines and AI systems rather than directly boosting rankings.

Read: Schema markup that actually matters (and what to skip) →

Search intent SEO & AI Search

Search intent is the goal behind a search: to learn something, find a specific site, compare options or buy. Matching a page to the dominant intent of a query matters more than matching its exact keywords.

Read: AI SEO vs traditional SEO: what actually changes →

Server-side tracking Analytics

Server-side tracking sends measurement data from your own server to analytics and ad platforms, rather than from the visitor's browser. It is more resilient to ad blockers and browser restrictions and gives you control over what data is shared.

Read: Server-side tracking: do you actually need it? →

Smart Bidding Paid Media

Smart Bidding is Google Ads' set of automated bid strategies, such as Target CPA and Target ROAS, that set a bid for each auction using conversion data. It needs accurate conversion tracking and enough conversion volume to learn from.

Read: Google Ads campaign structure that actually scales →

Structured output AI & Automation

Structured output is AI-generated data returned in a fixed, machine-readable format, usually JSON that matches a defined schema. It is what makes an AI step safe to place in the middle of an automation, because the next step can rely on the fields being present.

Read: Building your first AI agent: what it is and what it isn't →

System prompt AI & Automation

A system prompt is the standing instruction that sets an AI model's role, rules and boundaries for a whole conversation or application, separate from each user message. In a chatbot or voice agent it is where permitted actions and escalation rules belong.

Read: Prompt engineering for marketers: the parts that actually matter →

T

Token AI & Automation

A token is the unit AI models read and are billed in, roughly three quarters of an English word. Input and output tokens are priced separately, and output is usually several times more expensive.

Read: AI API costs explained: how to estimate before you build →

U

UTM parameters Analytics

UTM parameters are tags added to a link, such as utm_source and utm_campaign, that tell analytics tools where a visit came from. Consistent naming is what makes them useful; inconsistent tags fragment reporting.

Read: Setting up GA4 conversion tracking properly →

V

View-through conversion Analytics

A view-through conversion is a sale or lead credited to an ad that the person saw but did not click. It inflates platform-reported results and should be reported separately from click-through conversions.

Read: Attribution: why every platform over-reports, and what to do about it →

W

Webhook AI & Automation

A webhook is an automatic message one app sends to a URL the moment something happens, such as a form submission. It is the trigger behind most marketing automations because it delivers data instantly instead of waiting to be checked.

Read: Webhooks explained for marketers (no code required) →

Workflow automation AI & Automation

Workflow automation connects apps so that a repeatable process, such as routing a lead, sending a follow-up or building a report, runs without manual steps. Tools like Make.com and n8n handle the plumbing; the hard part is mapping the real process and its exceptions first.

Read: Make.com vs n8n: which one to build on →
Missing a term?

Ask, and I'll add it

If a term on this site or in your own work isn't here, send it through. For ready-made prompts that use these concepts, see the prompt library.