Expressions

The expression language is used throughout pipeline configurations for filter conditions, computed fields, and conditional mappings. It provides a SQL-like syntax for referencing event fields, applying operators, and calling functions.

Field References

Fields are referenced using dot notation starting from the event root:

Reference	Description
`event.type`	Event type (`track`, `page`, `identify`, `group`)
`event.event`	Event name (e.g. `"Purchase"`)
`event.userId`	Known user ID
`event.anonymousId`	Anonymous ID
`event.properties.revenue`	Event property
`event.traits.email`	User trait
`event.context.page.url`	Page URL from context
`event.context.geo.country`	Enriched geolocation
`event.context.device.browser`	Parsed device data
`event.context.session.id`	Session identifier
`event.vendorIds.ga4_client_id`	Vendor-specific identifier
`event.clickIds.gclid`	Captured click ID

Array Access

Access array elements by index or iterate with wildcards:

event.properties.products[0].name      -- first product name
event.properties.products[*].product_id -- all product IDs (returns array)

Nested Fields

There is no depth limit on nesting:

event.context.campaign.source
event.properties.items[0].variant.color

Operators

Comparison Operators

Operator	Description	Example
`=`	Equal	`event.type = 'track'`
`!=`	Not equal	`event.type != 'identify'`
`>`	Greater than	`event.properties.revenue > 100`
`<`	Less than	`event.properties.quantity < 10`
`>=`	Greater than or equal	`event.properties.revenue >= 50`
`<=`	Less than or equal	`event.properties.quantity <= 100`

Logical Operators

Operator	Description	Example
`AND`	Logical AND	`event.type = 'track' AND event.event = 'Purchase'`
`OR`	Logical OR	`event.event = 'Purchase' OR event.event = 'Checkout'`
`NOT`	Logical NOT	`NOT event.properties.is_test`

Set Operators

Operator	Description	Example
`IN`	Value in set	`event.event IN ('Purchase', 'Add to Cart', 'Checkout')`
`NOT IN`	Value not in set	`event.context.geo.country NOT IN ('CN', 'RU')`

Pattern Matching

Operator	Description	Example
`LIKE`	SQL-like pattern match (`%` wildcard)	`event.context.page.url LIKE '%/checkout%'`
`NOT LIKE`	Negated pattern match	`event.context.userAgent NOT LIKE '%bot%'`

Operator Precedence

From highest to lowest:

NOT
Comparison operators (=, !=, >, <, >=, <=)
IN, NOT IN, LIKE, NOT LIKE
AND
OR

Use parentheses to override precedence:

(event.event = 'Purchase' OR event.event = 'Refund') AND event.properties.revenue > 0

Functions

String Functions

Function	Description	Example
`LOWER(value)`	Convert to lowercase	`LOWER(event.traits.email)`
`UPPER(value)`	Convert to uppercase	`UPPER(event.properties.currency)`
`TRIM(value)`	Remove leading/trailing whitespace	`TRIM(event.traits.name)`
`CONCAT(a, b, ...)`	Concatenate strings	`CONCAT(event.traits.firstName, ' ', event.traits.lastName)`
`SUBSTRING(value, start, length)`	Extract substring	`SUBSTRING(event.properties.sku, 0, 3)`
`REPLACE(value, search, replace)`	Replace occurrences	`REPLACE(event.event, ' ', '_')`
`LENGTH(value)`	String length	`LENGTH(event.traits.email)`

Null Handling

Function	Description	Example
`COALESCE(a, b, ...)`	Return first non-null value	`COALESCE(event.userId, event.anonymousId, 'unknown')`
`IS_NULL(value)`	Check if null/missing	`IS_NULL(event.userId)`
`IS_NOT_NULL(value)`	Check if present	`IS_NOT_NULL(event.properties.revenue)`

Hash Functions

Function	Description	Example
`HASH_SHA256(value)`	SHA-256 hash (hex)	`HASH_SHA256(LOWER(TRIM(event.traits.email)))`
`HASH_MD5(value)`	MD5 hash (hex)	`HASH_MD5(event.traits.email)`

Date/Time Functions

Function	Description	Example
`NOW()`	Current Unix timestamp (seconds)	`NOW()`
`TO_UNIX(value)`	Convert ISO 8601 to Unix seconds	`TO_UNIX(event.timestamp)`
`TO_UNIX_MS(value)`	Convert ISO 8601 to Unix milliseconds	`TO_UNIX_MS(event.timestamp)`
`TO_ISO(value)`	Convert Unix seconds to ISO 8601	`TO_ISO(event.properties.created_at)`

Numeric Functions

Function	Description	Example
`ROUND(value, decimals)`	Round to N decimals	`ROUND(event.properties.revenue, 2)`
`FLOOR(value)`	Round down	`FLOOR(event.properties.revenue)`
`CEIL(value)`	Round up	`CEIL(event.properties.revenue)`
`ABS(value)`	Absolute value	`ABS(event.properties.amount)`
`MIN(a, b)`	Minimum of two values	`MIN(event.properties.quantity, 100)`
`MAX(a, b)`	Maximum of two values	`MAX(event.properties.revenue, 0)`

Type Functions

Function	Description	Example
`TO_STRING(value)`	Convert to string	`TO_STRING(event.properties.quantity)`
`TO_NUMBER(value)`	Convert to number	`TO_NUMBER(event.properties.price)`
`TO_BOOLEAN(value)`	Convert to boolean	`TO_BOOLEAN(event.properties.is_active)`
`TYPE_OF(value)`	Return the type as a string	`TYPE_OF(event.properties.revenue)`

Examples

Filter: Only Purchase Events

- type: filter
  condition: "event.type = 'track' AND event.event = 'Purchase'"

Filter: Events with Revenue

- type: filter
  condition: "IS_NOT_NULL(event.properties.revenue) AND event.properties.revenue > 0"

Filter: Specific Countries

- type: filter
  condition: "event.context.geo.country IN ('US', 'CA', 'GB', 'AU')"

Filter: URL Pattern

- type: filter
  condition: "event.context.page.url LIKE '%/checkout%' OR event.context.page.url LIKE '%/purchase%'"

Compute: Full Name

- type: compute
  fields:
    user_data.name:
      expression: "CONCAT(COALESCE(event.traits.firstName, ''), ' ', COALESCE(event.traits.lastName, ''))"
      type: string

Compute: Hashed Email (for Meta CAPI)

- type: compute
  fields:
    user_data.em:
      expression: "HASH_SHA256(LOWER(TRIM(event.traits.email)))"
      type: string

Compute: Revenue in Cents to Dollars

- type: compute
  fields:
    custom_data.value:
      expression: "ROUND(event.properties.amount_cents / 100, 2)"
      type: number

Compute: Conditional Value

- type: compute
  fields:
    events[0].params.value:
      expression: "COALESCE(event.properties.revenue, event.properties.total, event.properties.amount, 0)"
      type: number

Compute: UTM Source Normalisation

- type: compute
  fields:
    events[0].params.campaign_source:
      expression: "LOWER(COALESCE(event.context.campaign.source, 'direct'))"
      type: string

Null Behaviour

Expressions follow SQL-style null semantics:

Expression	Result
`NULL = 'value'`	`false`
`NULL != 'value'`	`false`
`NULL > 0`	`false`
`NULL AND true`	`false`
`NULL OR true`	`true`
`NOT NULL`	`NULL`
`COALESCE(NULL, 'fallback')`	`'fallback'`

When a field reference points to a field that does not exist on the event, it evaluates to NULL. Use COALESCE() or IS_NULL() / IS_NOT_NULL() to handle missing fields gracefully.

Type Coercion

The expression engine performs automatic type coercion in these cases:

Context	Coercion
Comparison with string literal	Value is coerced to string
Comparison with number literal	Value is coerced to number
Arithmetic operations	Both operands coerced to number
`CONCAT()` arguments	All arguments coerced to string
Boolean context (filter conditions)	Truthy/falsy evaluation

If coercion fails (e.g. converting "abc" to a number), the expression evaluates to NULL.

Pipeline Configuration Custom Code