Filter values
Keep filters from matching every record when their comparison value turns out empty.
A filter, or a search condition inside an action, compares a field with a value. Tape builds that value when the automation runs, from fixed input and references: a field, a variable or the output of an earlier action. A filter that looks complete can still end up with an empty value for one record.
An empty filter value matches every record
If it is empty for one record when the automation runs, the condition is true for every record. Under the default and group, the automation then runs its actions on every record in scope, without an error. Before a bulk action like "update all" or "delete all", this can change your whole dataset and use credits for every record.
400 when you save a filter?Create and update in the Automation API reject a comparison that is empty or of the wrong type, and name the condition:
- a value whose
typedoesn't match the field, for exampletexton a number field - an empty comparison value, for example
{ "type": "text", "text": [] } - the wrong number of values
The error explains the risk. To check for an empty field on purpose, use is_empty or is_not_empty.
How a value ends up empty
- A referenced value is empty. The field, variable or earlier output is empty for this record. It passes the save check, because the reference exists.
- The value channel doesn't match the field. Tape catches this on save and returns
400. Tape picks the channel from the field type: number and unique ID usenumber, date fields usedate, category, status, relation and user useids, text usestext. See Match the value type to the field for the channel rules. - The automation is older than the check. Automations saved before this check, or in the automation editor, can hold an empty comparison. Create and update don't check existing automations again.
How to avoid it
- Check the source first. If the value comes from a reference that can be empty, put the filter or action inside a
conditionalthat checks the source, or set a fallback value. - Simulate before you activate. A simulation runs the actions on real records without changing data in Tape, so a search condition inside an action that matches too much shows up early. Requests to other services still go out. A simulation skips the filter of the automation itself.
- To match every record on purpose, remove the condition, or send
filter: null. A filter without conditions runs for every record. An empty top-level group returns400.
Why validate does not flag an empty value
GET /v1/automation/{id}/validate returns valid: true for a condition with an empty comparison value, because the definition is valid. validate checks meaning only: it doesn't run the save checks of create and update, and can't know that a reference becomes empty for some record.
validate catches missing references, which is a different problem:
filter_variable_missing: the subject of a condition points to a field that no longer exists.filter_value_variable_missing: the value points to a deleted field or variable.
Neither code flags an empty comparison. The code filter_value_missing is reserved and no longer raised.
Next steps
- Run & simulate: test the actions of an automation without changing data in Tape.
- Filters: conditions, operators and value channels.
- Execution types: how filters apply in each type of run.
- Errors: read failed runs.
- In the Help Center: My filter doesn't match