Filters

Filters let you isolate and analyze subsets of your data. You can use filters on any query type, including the Funnel Report, Cohort Report, People Search, and A/B Test Report.

They can be used to isolate data by specific events or properties. Furthermore, you can use filters to isolate data where a person did a certain event with a specific property or just property. Filters become even more powerful when you nest or chain them together using logical operators.

Filter object properties

Property Name	Required?	Description
`type`	Yes	Filter type: `event` `property`

`negate`	No	Set to `true` when you want to filter data by "has not done event" or "does not have property", instead of the default positive "has done" or "has property".

`with_property`	No	Set to `true` when you want to filter by events which have a specific property attached to it.

`while_property`	No	Set to `true` when you want to filter by events where the person has a specific property.

`event`	Yes, if the `type` is `event`.	The index of the event.

`property`	Yes, if one of the following: `type` is `event` `with_property` is `true` `while_property` is `true`	The index of the property.

`first_ever`	Optional if the filter `type` is `event`	Set to `true` when you want to filter data where the person did the the event for the first time.

`frequency_occurrence`	Optional if the filter `type` is `event`.	Set to `at_least`, `at_most`, or `exactly` if you want to filter by the number of times an event was done.

`frequency_value`	Yes, if both of following are true: `type` is `event` `frequency_occurrence` is set	The value used to calculate the frequency occurrence. For example, at least `1` or exactly `6`.

`comparison_mode`	Optional if one of the following: `type` is `property` `with_property` is `true` `while_property` is `true`	The comparison mode to use when filtering by properties, or events with properties. Use comparison modes to filter for specific property values. See Property Filter Comparison Modes.

`comparison_string`	Required if `comparison_mode` is not `in_range`, `any_value`, or `empty`.	The string to use when comparing property values.

`comparison_range_min`	Required when the `comparison_mode` is `in_range`.	The minimum value when comparing a property that should be within a range.

`comparison_range_max`	Required when the `comparison_mode` is `in_range`.	The maximum value when comparing a property that should be within a range.

`date_range`	No	A Date Range object.

Property Filter Comparison Modes

Property Value Type	Acceptable Comparison Modes
Text values	`"contains"`, `"begins_with"`, `"ends_with"`
Numeric values	`"greater_than"`, `"less_than"`, `"greater_than_or_equal_to"`, `"less_than_or_equal_to"`, `"in_range"`
Any values	`"any_value"`, `"empty"`, `"equals"`

Logical Operators

More complex queries will require that multiple filters be chained together as is the case for constructing:

All who did/have any of these events/properties (type: "or")
All who did/have all of these events/properties (type: "and")
All who did/have all of these events/properties in order (type: "then")

Here's how the filter object looks when using logical operators:

Property Name	Required?	Description
`type`	Yes	One of the following: `or` `and` `then`

`operands`	Yes	An array of Filters. This can also include nested logical operators, see Nesting Filters .

`date_range`	No	A Date Range object.

Nesting Logical Operators

Filters can be nested to abitrary depths by placing Logical Operators within Logical Operators. To visualize, imagine a tree, where Logical Operators can be either AND/OR/THEN and filters can be Event, Property, or EventWithProperty:

// Expression Tree
             AND
          /     \
        OR       OR
       /  \      /  \
  Event  Event  AND  Property
               /  \
        Property   EventWithProperty