n8n Split Out Node Examples: One List, Many Items

Why do I need Split Out?

n8n runs each node once per item. If one item holds a list of seven tasks, the next node still sees one item, and a sheet node would write one row.

In the lesson, Message a model returns 2 items, one per meeting. Each holds a list called tasks: 4 for the first meeting and 3 for the second. Each meeting is still one item that holds a list, and that's what Split Out fixes. The glossary has a short Split Out node entry.

How do I set up Split Out?

Add it after the node that returns the list, then give it the path to that list.

1. Click the + after Message a model, type Split Out and click it. It has no sub-menu, so it opens straight away.
2. Click Fields To Split Out and type the path: output[0].content[0].text.tasks. In n8n you can instead drag tasks from the left panel.
3. Click Execute step. The OUTPUT shows 7 items, each with task, owner and due.

Two meetings became seven rows of work, and the Google Sheets Append row in sheet node after it writes one row per task.

How do I find the right path?

Look at the input on the left and follow the nesting from the top down, joining names with dots and using [0] for the first entry of a list.

Dragging the field from the INPUT panel writes the path for you, which avoids typos.

What does the "o" field error mean?

If you see an error about a field called "o", n8n ran the node while the path was half typed: it tried to split a field named "o", the first letter of "output". Click Execute step again once the full path is in. The lesson calls this out because it is easy to click too early.

What about other split errors?

• The field isn't a list. If the path points at text, the AI probably returned JSON as plain text. In the lesson, setting the OpenAI node's Output Format to JSON Object makes n8n parse the answer so tasks arrives as a real list. See n8n OpenAI JSON Output.
• The field doesn't exist on some items. One meeting with no tasks may return no list. Check the AI's output for every item.

Can I keep fields from the parent item?

Yes. Under Include, choose to keep other fields or selected fields, so each split item can still carry, say, the meeting name. Without it, each new item only has the fields from the list entry.

Is there a way to skip Split Out?

If the list comes from somewhere you control, you can produce one item per entry from the start. In the Meeting to Action lesson, ChatGPT returns a JSON array that you paste as mock data, and n8n makes one item per object directly: 4 items, no Split Out needed.

Key takeaways

• Split Out turns a list inside one item into separate items.
• Fields To Split Out is the path to the list, like output[0].content[0].text.tasks.
• An error about a field called "o" means you ran it before the path was complete.
• The list must be real JSON, so set the AI's output format to JSON.

Frequently asked questions

What does the Split Out node do in n8n?

It takes an array field from each incoming item and outputs one item per array entry, so later nodes run once per entry.

What is the difference between Split Out and Loop Over Items?

Split Out creates more items from a list. Loop Over Items processes existing items in batches. Most office workflows only need Split Out, because n8n already runs nodes once per item.

Why does Split Out say the field is not an array?

The value at that path is text or an object, not a list. Usually the AI answer wasn't parsed as JSON, or the path stops one level too early.

Which course is this in?

AI for Operations Teams and AI for Meetings.