Workflow Fundamentals

An n8n workflow consists of nodes connected by connections. Each node performs a specific action — fetch data, transform, send, or branch. In this lesson, you'll learn the core concepts.

Nodes — The Building Blocks

Nodes are the basic units of every workflow. There are three main categories:

Trigger Nodes

Start the workflow. Without a trigger, nothing happens.

Action Nodes

Perform the actual work:

• HTTP Request: Call any API
• Database nodes: Query PostgreSQL, MySQL, MongoDB
• App nodes: Slack, Google Sheets, HubSpot, Jira, etc.
• Code node: Execute JavaScript or Python
• AI nodes: LLM calls, embeddings, text classification

Logic Nodes (Flow Control)

Control the workflow flow:

• IF: Conditional branching
• Switch: Multiple branching
• Merge: Combine data
• Loop Over Items: Iterate over lists
• Wait: Time delay or wait for event

Connections and Data Flow

Nodes are connected by connections. Data flows as items — JSON objects passed from node to node:

[
  { "json": { "name": "Max Mueller", "email": "max@company.com", "score": 85 } },
  { "json": { "name": "Anna Schmidt", "email": "anna@company.com", "score": 42 } }
]

Each node receives items, processes them, and outputs new items.

Expressions — Dynamic Values

Expressions allow access to data from previous nodes:

{{ $json.name }}                    → Current item value
{{ $('HTTP Request').item.json.id }} → Value from a specific node
{{ $now.toISO() }}                  → Current timestamp
{{ $json.score > 70 ? 'Good' : 'Bad' }} → Conditional logic

Error Handling

Robust workflows need error handling. n8n offers three approaches:

1. Error Trigger Workflow

A separate workflow triggered on errors in other workflows — ideal for centralized alerting (Slack, email).

2. Retry on Fail

Nodes can automatically retry on failure:

• Retry Count: Number of attempts (e.g., 3)
• Wait Between: Wait time between attempts (e.g., 5 seconds)

3. Continue on Fail

The workflow continues even if a node fails. The error is passed as output and can be handled in the next node.

Execution Modes

Practical tip: Enable error handling from the start. Create a central error workflow that reports failures to Slack or via email. This way, you'll know immediately when a production workflow fails.