Your first automation

T05-L01 · Automation · Level 1 User · 24 minutes

This mini-book is for one small workflow that helps with your own week. Use a test workspace, synthetic data, and accounts approved by your lab or company. You remain responsible for the input, the action, and stopping the workflow.

2. Monday's six-copy routine

Every Monday, you copy the same six details from one place to another. In a lab, a colleague adds a file to a shared test folder, and you type its filename, sample code, owner, file type, location, and arrival time into a tracking sheet. In a company, someone submits a request form, and you copy its request ID, topic, summary, team, status, and submission time into a sheet before confirming receipt.

This week, you miss one row. Nobody notices until Thursday. The file is present but absent from the tracker, or the requester submits the form again because no acknowledgement was visible.

You do not need an intelligent agent. You need a narrow automation: one event starts one predictable action. You will build it with invented records, test a success, create one safe failure on purpose, inspect the run history, and find the off switch before enabling anything.

3. After this you can

• Build one trigger-and-action workflow from end to end.
• Map six synthetic source fields into a test spreadsheet row.
• Test a normal run and a deliberately failed run before activation.
• Explain where the workflow runs when your laptop is closed.
• Stop the workflow with the correct off switch.

4. Prerequisites

• T01-L03 · The simplest thing that works.
• An approved n8n test workspace.
• A test folder or form and a test spreadsheet that contain no real records.
• Permission to connect those test services. Use a dedicated test account if your organisation provides one.
• About 45 minutes for the independent exercise.

Do not use participant details, unpublished findings, real sample records, customer information, employee information, contracts, credentials, or production exports. Never paste an API key or password into a normal workflow field. If you cannot obtain an approved test workspace, read the build and reproduce the mapping on paper; do not connect a personal account as a shortcut.

5. The idea in one page

An automation begins with a trigger and continues with an action.

• The trigger is the exact event that starts a run: for example, a new file in one named test folder or a new response to one named test form.
• The action is the bounded change the run attempts: here, adding one row to one test sheet.
• Field mapping says which trigger value enters which action field. A file's name belongs in item_name; its timestamp does not.
• The run history records what happened in each attempt. Open it to inspect the trigger input, mapped values, action result, time, and error. A green status alone does not prove that the right values reached the right columns.

Write the contract before building:

WHEN: one new synthetic item appears in the named test source
COPY: item_id, item_name, owner, category, status, event_time
TO: one new row in the named test sheet
NEVER: message a real person, overwrite a row, or delete anything
OWNER: [your name or role]
STOP: [the exact control that disables the workflow]

A hosted automation service runs on the provider's systems after activation, so closing your laptop usually does not stop it. A self-hosted tool runs only while its host and required services are available. Closing a laptop that hosts the tool may stop runs; closing a laptop used only to edit a server-hosted tool does not. Confirm the actual deployment rather than guessing from the browser window.

Find three n8n controls before connecting the trigger: Execute Workflow, the Executions history, and the workflow's published/unpublished control. Keeping the workflow unpublished prevents production trigger events from starting it; it does not undo an action that already succeeded.

Tool-choice comparison. Use n8n when you want deeper control and a path to self-hosting, Make when a visual scenario is the useful middle ground, or Zapier when broad connector coverage is the deciding need. The implementation below covers n8n only, in an approved test workspace; it does not claim that its steps or labels apply to Make or Zapier.

The first and last stages are this lesson's trigger and action; the middle stages show later options, not requirements. Course-authored SVG with synthetic labels; reviewed on 4 September 2026 for privacy and product-neutral staleness.

6. The worked example: build it in n8n

The Lab and Company versions use the same skeleton:

named test source -> new-item trigger -> map six fields -> add one sheet row

Create a Google spreadsheet named First automation - synthetic test with one sheet and this header row:

item_id,item_name,owner,category,status,event_time

Keep the n8n workflow unpublished while building. For the Lab framing, use a Google Drive Trigger node; for the Company framing, use an n8n Form Trigger node. In both framings, connect that single trigger directly to one Google Sheets action node. Do not add an Edit Fields, notification, acknowledgement, or error-handling node: this exercise deliberately preserves one trigger and one action.

Build the shared skeleton

1. Name the workflow. Use First automation - synthetic log. Add a description that names the source, destination, owner, and the control used to unpublish it.
2. Configure one trigger. Add the n8n trigger specified in your framing below. Restrict it to the named test folder or configure only the named test form. Do not point it at a production source.
3. Load one synthetic event. Create the Lab file or submit the Company form described below. Execute the trigger in test mode and inspect every output field.
4. Configure the one action. Add Google Sheets. Set Resource to Sheet Within Document and Operation to Append Row. Under Document, use From list to select First automation - synthetic test; select its only sheet under Sheet.
5. Map six values. Set Mapping Column Mode to Map Each Column Manually and map the six trigger values to Values to Send. Expressions inside these fields are part of this Google Sheets node, not extra actions. Do not type an example value into a field that should change on every run.
6. Run once. Select Execute Workflow, supply one event through the trigger, open the new row, and compare all six cells with the source event.
7. Inspect history. Open the successful execution in Executions. Check the trigger output, Google Sheets input and output, timestamp, and status.
8. Keep it unpublished. Do not publish it. First complete the deliberate failure test below.

Lab framing: a file arrives in a shared folder

Create a test folder called Northstar synthetic arrivals. Add a plain-text file named:

SYN-104__assay-notes__mira__txt.txt

The file contains only this invented text:

Synthetic training file. No sample, participant, or research data.

In the Google Drive Trigger, choose the event for a file created in a specific folder and select Northstar synthetic arrivals. n8n's Google Drive Trigger returns the latest matching event during a manual execution, so add the intended test file immediately before selecting Execute Workflow, then verify its name in the trigger output. In the Google Sheets node, derive item_id with an n8n expression that splits the current filename on __ and takes the first segment; do not fix SYN-104 as text. If the trigger does not supply one of the other five values, use a clearly labelled fixed synthetic value in that mapped column only for this exercise.

After the run, the row must identify the same file and preserve the trigger timestamp. Do not substitute the time at which you happen to inspect the sheet. A later review should be able to connect the row to the source event.

Company framing: a form request is received

Configure the n8n Form Trigger with the title Harborlight synthetic requests and field names request_id, topic, summary, and team. Use its Test URL by selecting Execute Workflow, then submit this invented response:

Request ID: DEMO-208
Topic: delivery
Summary: Please confirm the fictional sample order window.
Team: Operations Demo

Set the Form Trigger to respond when the form is submitted and display generic response text such as Synthetic request received. That built-in n8n form response is the acknowledgement; it is not a second workflow action and it sends nothing to a person. Map the trigger output into the same Google Sheets node and sheet structure.

Check the on-screen acknowledgement immediately after submission. Then verify that the row says DEMO-208, not an ID remembered from an earlier test. This is still one automation trigger and one action: the form acknowledges; the workflow logs.

Produce a safe failed run

Now make the n8n Google Sheets action fail without exposing or changing real data. This uses the node's documented manual document-ID choice: Document can be selected By ID instead of From list.

1. Confirm that the successful synthetic run is visible in n8n's Executions history.
2. Open the Google Sheets node and record that Document is From list with First automation - synthetic test selected. Change only the Document selection mode to By ID and temporarily enter SYNTHETIC_INVALID_DOCUMENT_ID_T05L01. This invented value is deliberately not a real Google spreadsheet ID. Do not change the Google credential, trigger, sheet mapping, or service permissions.
3. Save the unpublished workflow. The change is reversible because you recorded the valid From list selection and changed no data or permissions.
4. Create one new synthetic source event with a different ID. For Lab, add SYN-FAIL-105__assay-notes__mira__txt.txt to the same test folder; for Company, submit a new synthetic form response whose request ID is DEMO-FAIL-209.
5. Test the trigger and inspect the add-row action input before execution. It must show item_id as SYN-FAIL-105 for Lab or DEMO-FAIL-209 for Company, not the successful event's ID. If it does not, repair the source-field mapping before continuing; do not replace it with fixed example text.
6. Select Execute Workflow once. The trigger should supply that event, and the Google Sheets node should fail because its Document value is an invalid synthetic spreadsheet ID.
7. Open the failed record in Executions. Confirm that the trigger output and Google Sheets input show the failed event's distinct ID, and that n8n marks the Google Sheets node as failed with a missing, inaccessible, or invalid-document error. Do not retry repeatedly.
8. Reopen the Google Sheets node, change Document from By ID back to From list, restore First automation - synthetic test, save, and leave the workflow unpublished. Confirm the destination is restored, but do not run the failed event again.

The two run records now answer different questions. The success shows that the trigger, mapping, and action can complete. The failure shows where errors appear and proves that you can inspect them. Neither result authorises live data or unattended use.

7. What goes wrong

You enable before testing

Symptom: real events start creating rows while fields are still unmapped or pointed at example values.

Fix: leave the workflow disabled, test with synthetic events, compare every cell, and activate only after an authorised review.

The trigger listens to everything

Symptom: every file in a drive or every form in a workspace starts a run.

Fix: select one named test source. Reopen the trigger settings and state the exact event that may start it.

Example text replaces mapped data

Symptom: every row contains DEMO-208 or the same timestamp, even when the source event changes.

Fix: inspect the trigger output and map its field token into the action. Test with a second ID so copied constants are obvious.

A green run hides a wrong row

Symptom: history says successful, but the owner appears under category or the timestamp is the inspection time.

Fix: compare the source event, action input, and destination row field by field. Success means the operation completed, not that your mapping was correct.

Failure is silent

Symptom: rows stop appearing, but nobody checks the run history and no approved failure notification exists.

Fix: name the owner and a routine for checking failed runs before activation. Add notifications only through an approved channel and test process.

Nobody else can stop it

Symptom: the workflow is built under a personal account, and the owner is unavailable when it misbehaves.

Fix: use an organisational test workspace where possible, record the owner and off switch, and confirm that an authorised backup can disable it before live use.

8. Do it yourself: 45 minutes

Minutes 0–5: choose the Lab or Company framing. Write the six-line contract from section 5. Locate Execute Workflow, Executions, and the n8n published/unpublished control.

Minutes 5–12: create the synthetic source and destination. Confirm that neither contains real personal, customer, participant, sample, or confidential information.

Minutes 12–25: build one specified n8n trigger and one Google Sheets Append Row action. Load the synthetic event, map all six fields, and keep the workflow unpublished.

Minutes 25–32: run the normal case. Compare the six destination cells with the source and inspect the action input and result in run history. Repair mapping errors one at a time.

Minutes 32–39: change the Google Sheets Document mode from From list to By ID and enter the reversible test value SYNTHETIC_INVALID_DOCUMENT_ID_T05L01. Trigger SYN-FAIL-105__assay-notes__mira__txt.txt for Lab or a form response with request ID DEMO-FAIL-209 for Company. Confirm that distinct ID in the failed node input, inspect the document error, then restore From list and the recorded test spreadsheet immediately.

Minutes 39–43: write the workflow owner and exact off-switch location in its description. Confirm whether it is cloud-hosted or self-hosted and what closing your laptop would do.

Minutes 43–45: leave the workflow unpublished. In Executions, arrange the history list so the successful and deliberately failed records for this workflow are visible together. The detailed IDs, open error, restored destination, and unpublished state are exercise checks; they do not all have to fit in the evidence screenshot.

9. Exit check

Deliver exactly one artifact: one evidence screenshot of n8n run history showing one successful and one deliberately failed run for the same workflow.

It passes when the one screenshot lets another person identify the same n8n workflow and see one successful status and one deliberately failed status. It does not need to prove the restored configuration, unpublished state, detailed synthetic IDs, or an open error panel; those are exercise checks. Crop out account addresses, tokens, secrets, and unrelated records. The workflow itself is the exercise; do not submit a second artifact.

10. Rule to remember

Know how to stop it before you start it.

11. Further reading & tools

• Taught: T01-L03 · The simplest thing that works — confirms when a fixed automation is simpler than chat, an app, or an agent.
• Taught: Automation tools — practises trigger, field mapping, action boundary, and run-history inspection with synthetic input.
• Taught: n8n — introduces manual executions and inspectable workflow data before credentials or external actions.
• Catalogued: n8n: Build your first workflow (opens in a new tab) — primary documentation for triggers, nodes, executions, and workflow activation.
• Catalogued: n8n: Google Sheets Append Row (opens in a new tab) — documents Document selection From list, By URL, or By ID, plus manual column mapping.
• Catalogued: Tools index — compare the locally catalogued automation tools, then confirm current capabilities and approval with the primary vendor documentation and your organisation.