Summarize items

Group your items by one or more fields, then count, add up, average or list values in each group. It is the pivot table of a spreadsheet, as a step.

Not the AI summary

This step does arithmetic on items. To summarise a text with AI, use Summarize (on-device) or the Summarization Chain.

What it does

Summarize items receives every item at once.

• Group by: one or more fields (dot paths such as customer.country). Leave it empty for one summary of all items.
• Summaries: one card per calculation. Each card has a Calculation, a Field and an optional Output name:

Calculation	What it gives	Default output name
Count items	How many items are in the group (no field needed)	count
Count unique	How many different values the field has	count_unique_email
Sum	The total of the field	sum_price
Average	The mean of the field	avg_price
Min / Max	The smallest / largest value (ISO dates compare as dates)	min_price / max_price
Join as text	The values joined with a Separator (default , )	joined_name
Collect as a list	Every value, as a list	list_name

Groups appear in the order their first item arrived. Values are compared exactly: 1 and "1" are two different groups. Values that are not numbers are skipped by Sum and Average, with one warning that names the field and the items, for example 3 values in "price" weren't numbers and were skipped (items 4, 9, 12). After a run, the step shows how many groups it made.

When to use it

• Sales per region, tickets per status, emails per sender.
• One total line at the end of a list (leave Group by empty).
• Before a chart or a report step, to turn many rows into a few.

Inputs and settings

Réglage	Type	Ce que cela fait
Group by	Tags	Leave empty for one summary of all items.
Summaries	List of entries	Par défaut : [{"op":"count"}]
Summaries (each entry) › Calculation	Choice	Options: Count items, Count unique, Sum, Average, Min, Max, Join as text, Collect as a list. Par défaut : Count items
Summaries (each entry) › Field	Text	Dot path, e.g. price or customer.email. Count without a field counts items.
Summaries (each entry) › Output name	Text	Defaults to e.g. sum_price, count, count_unique_email.
Summaries (each entry) › Separator	Text
Output	Choice	Options: One item per group, One item with all groups. Par défaut : One item per group
Options	Group
Options › Skip empty values	Toggle	Par défaut : On
Options › When a field is missing	Choice	Options: Ignore the item for that summary, Stop the step. Par défaut : Ignore the item for that summary

Set Output to choose the shape:

• One item per group (default): each group is an item with its group fields and its summaries.
• One item with all groups: a single item { groups: [...] }, handy for one message or one row.

Under Options, Skip empty values (on by default) leaves empty values out, and When a field is missing either ignores that item for the summary or stops the step.

Because the step knows its output fields from the settings, the next step’s data picker shows them before you run anything.

Outputs

• Output

Group by region · Sum of price · Count items

[
  { "region": "EU", "sum_price": 1250, "count": 12 },
  { "region": "US", "sum_price": 980, "count": 9 }
]

Dependencies and credentials

None. This node needs no connection and has no slots for other nodes.

None.

Example workflow

1. HTTP Request gets the to-dos of a project, one item each.
2. Filter keeps the completed ones.
3. Summarize items, Group by userId, Count items: how many each person finished.

Common issues

• One group per item: the Group by field is different on every item (an id or a date with time). Group by a coarser field.
• The sum is lower than expected: some values are text such as "12,50". Check the warning in the run log, and fix the values with Edit Fields first.
• A field is missing in the output: give the card a Field; only Count items works without one.

Related nodes

• Aggregate: gather fields from all items into lists, without grouping.
• Remove Duplicates: keep one item per value.
• JavaScript Code: any other calculation.