Glossary — plain-language definitions

Keep this open while you read the kit. Terms are grouped by where you will meet them.

Business terms

Agency — a service business that does work for clients. Here: builds and maintains automations.
Productized service — a service with a fixed name, scope, price and process, sold like a product.
Retainer / MRR (monthly recurring revenue) — a fixed monthly fee for ongoing service.
Setup fee — a one-time fee for building and launching an automation.
White-label — you build, another company sells it under its own brand.
Scope — the written list of what is included and what is not.
Change request — a written ask to add or alter something outside the scope, usually priced separately.
SOW (statement of work) — a document that lists deliverables, timeline, responsibilities and price.
SOP (standard operating procedure) — a step-by-step checklist for doing a task the same way every time.
Discovery — the conversation and research phase where you learn the client's process and pain before proposing.
Upsell / expansion — selling an additional automation or plan to an existing client.
Churn — clients cancelling. The enemy of MRR.
ROI (return on investment) — value gained divided by cost. Used to justify your price.
Loaded hourly cost — an employee's wage plus taxes, benefits and overhead, per hour. Usually 1.25–1.4× salary.

Technical terms

Automation / workflow — a chain of steps that runs automatically when something happens.
Trigger — the event that starts a workflow: a form is submitted, an email arrives, it is 9:00 on Monday.
Node — one step in an n8n/Make workflow (a box on the canvas). Each node does one job.
API (application programming interface) — a way for one software to talk to another by sending requests. Think of it as a waiter taking orders between two kitchens.
API key — a secret password that identifies you to an API. Treat it like a bank password.
Webhook — a URL that a service calls when something happens ("push" instead of "ask"). Your workflow listens at that URL.
Polling — the workflow checks a service every few minutes for new items. Used when a service has no webhook.
HTTP request — the basic way to call an API: a method (GET, POST), a URL, headers, and often a body.
JSON — the text format APIs use to exchange data: {"name": "Ana", "score": 82}. Keys and values in braces.
Structured data — data with a fixed shape (fields, types). The opposite of free text.
Expression — in n8n, a small piece of code inside {{ }} that pulls a value from a previous step, e.g. {{ $json.email }}.
Credential — a stored login (OAuth connection or API key) that n8n uses to talk to a service. Stored encrypted, never inside the workflow file.
OAuth — the "Sign in with Google" style of authorisation. You click Allow; no password is shared.
LLM (large language model) — the AI models that read and write text: GPT (OpenAI), Claude (Anthropic), Gemini (Google).
Prompt — the instructions you give a model. System prompt = standing instructions (role, rules, format); user prompt = the task and the data.
Token — the unit models count text in; roughly 3/4 of a word in English. Pricing is per million tokens.
Temperature — a setting from 0 to 1 (or 2): low = predictable, high = creative. Use low for business automations.
Context / context window — everything the model can see at once (prompt + data). It has a maximum size.
Memory — for chat use-cases, storing previous messages so the model remembers the conversation. Automations usually have no memory; each run is fresh.
Hallucination — the model states something false with confidence. Mitigated by giving it the facts, constraining the output, and human review.
Prompt injection — when text the model reads (an email, a web page) contains instructions that hijack it ("ignore your rules and…"). Mitigated by separating data from instructions and limiting what the workflow can do.
RAG (retrieval-augmented generation) — giving the model the relevant documents at question time so it answers from your data instead of guessing.
Vector database — a database that finds "similar meaning" text; used in RAG. ADVANCED; not required for the core automations.
Rate limit — the maximum number of requests a service allows per minute/day. Exceeding it returns error 429.
Error handling — what the workflow does when a step fails: retry, skip, alert a human.
Human-in-the-loop (HITL) — a step where a person approves or edits before the workflow continues.
Idempotent — running the same step twice does not create a duplicate. Important for "send" and "create" actions.
Environment variable — a setting stored outside the workflow (e.g. a URL or threshold) so it can be changed without editing nodes.
Staging vs production — a test copy versus the live system. Test in staging (your own accounts) first.
Latency — how long a step takes. AI steps typically take 2–20 seconds.
CRM — the system where a business keeps its contacts and deals (HubSpot, Pipedrive, GoHighLevel, or just a Google Sheet).
Sticky note — a coloured note on the n8n canvas used for instructions. Every kit workflow has one.

Data protection terms

Personal data / PII — anything that identifies a person: name, email, phone, address, IP.
GDPR — the EU regulation on personal data. Principles: lawful basis, minimisation, purpose limitation, security, retention limits, individual rights.
CCPA/CPRA — California's consumer privacy laws, similar principles.
Data processor vs controller — the client decides why and how data is used (controller); you process it on their behalf (processor). A written agreement (DPA) usually governs this.
Data retention — how long data is kept before deletion. Automations should not keep more than needed.
Least privilege — every credential gets only the permissions it needs, nothing more.