n8n tutorial: your first workflow in 20 minutes (tested on n8n 2.45)
By Clément Lacaille (Tech-Bharat) · Published 31 October 2026 · 3 min read · how we build and test
First Workflows: 9 Step-by-Step Workshops — €19
PDF guide in English on this topic · 21 pages · October 2026 edition · 14-day guarantee
To build your first n8n workflow, add a manual trigger, call a public API with the HTTP Request node, extract three values with Edit Fields, branch with an If node, then swap the trigger for a Schedule Trigger and click Publish. This tutorial does it on a concrete example — getting warned if it rains tomorrow — which we built and ran on n8n 2.45.7 (official Docker image) on 31 October 2026. The outputs quoted are the ones n8n returned that day.

What you need first
An n8n 2.x instance: n8n Cloud, or n8n on your own machine (see install n8n locally). No third-party account or API key: the tutorial uses Open-Meteo, a public weather API, free for non-commercial use, with no key. Allow 20 minutes.
Two words to know: a node is one step of the workflow; an item is a unit of data passed from node to node. A node runs once per item it receives — remember that, it explains many surprises.
Step 1: call the API
- Create a workflow and add the Manual Trigger ("When clicking 'Execute workflow'"). It is for testing only.
- Add an HTTP Request node: method
GET, URLhttps://api.open-meteo.com/v1/forecast. - Turn on Send Query Parameters and add:
latitude=48.8566,longitude=2.3522(Paris — use your own city),daily=precipitation_sum,temperature_2m_max,timezone=Europe/Paris,forecast_days=2. - Click Execute workflow.
Result in our lab (excerpt):
{
"timezone": "Europe/Paris",
"daily": {
"time": ["2026-10-31", "2026-11-01"],
"precipitation_sum": [0, 0],
"temperature_2m_max": [15.7, 12.8]
}
}
The API does not return a list of days but three parallel lists. Tomorrow is at position 1 of each (counting from 0). Reading the exact shape of a response before building the next step is the first habit to adopt.
Step 2: extract and decide
- Add Edit Fields (Set), named "Tomorrow", with three fields:
date(String) ={{ $json.daily.time[1] }};rain_mm(Number) ={{ $json.daily.precipitation_sum[1] }};max_temp(Number) ={{ $json.daily.temperature_2m_max[1] }}. - Add an If node: value
{{ $json.rain_mm }}, operator Number › is greater than or equal to, value1. - Run it.
In our lab, "Tomorrow" returned the date 2026-11-01, 0 mm of rain and a maximum of 12.8 °C. The item went out through the If node's false output (no rain forecast). Connect whatever you like to the true output: an email, a Telegram message, a table row.

Step 3: schedule and publish
- Replace the Manual Trigger with a Schedule Trigger: Trigger Interval = Days, Days Between Triggers = 1, Trigger at Hour = 7am.
- In the workflow settings (… menu › Settings), check Timezone.
- Click Publish.
Since n8n 2.0 the editor saves your changes without touching the production version: only the last published version runs (what changes with Save and Publish). Runs show up in the Executions tab, with every node's data.
The 4 traps we ran into
| Trap | What happens | The fix |
|---|---|---|
| Time zone | Without a setting, a self-hosted instance uses GENERIC_TIMEZONE, which defaults to America/New_York: "7 a.m." fires at 7 a.m. New York time |
Set the workflow time zone, and GENERIC_TIMEZONE on the instance |
| Types | The If node compares strictly: the text "0" compared with a number raises a type error | Declare the field as Number, or turn on Convert types where required |
| Not publishing | A fix that is saved but not published never applies in production | Click Publish after each validated change |
| Silent failure | A failing API makes the execution fail without telling anyone | An error workflow (Error Trigger) set in the workflow settings |
For the last point, see handling errors with an Error Workflow: in our lab, an API answering 503 produced the message "Service unavailable - try again later or consider setting this node to retry automatically (in the node settings)", received by the error workflow.
What next
This workflow already holds the bricks of most automations: a trigger, an API call, a transformation, a decision. Next come processing a list (Filter, Aggregate), storage (Data Tables), receiving data (Webhook, form) and error handling. To debug a workflow that doesn't do what you expect, each node's JSON view remains your best tool.
FAQ
Frequently asked questions
Do I need to code to follow this n8n tutorial?
No. The workflow uses three one-line expressions such as {{ $json.daily.time[1] }} that read a field of the response; no Code node is needed.
Does this tutorial work on n8n Cloud?
Yes: Manual Trigger, HTTP Request, Edit Fields, If and Schedule Trigger exist on n8n Cloud as well as self-hosted. Only the GENERIC_TIMEZONE variable, which sets the instance's default time zone, is specific to self-hosting.
Why doesn't my scheduled workflow run?
In n8n 2.x a workflow only runs on its own once published with the Publish button; a change that is saved but not published never applies in production. Also check the workflow time zone: without a setting, a self-hosted instance uses America/New_York.
Why does the API return lists rather than a list of days?
Open-Meteo returns parallel arrays (time, precipitation_sum, temperature_2m_max): tomorrow's value is at index 1 of each. That is this API's shape; others return an array of objects, which n8n then splits automatically into one item per element.