Segmentation Query Language Reference

[This topic is pre-release documentation and is subject to change.]

Membership in static segments is defined by specifying either the Profile ID for each member or supplying a segmentation query; membership in a dynamic segment is defined by only by the latter. A segmentation query is a filtering expression that describes the desired membership set. Each expression is composed of one or more sub-expressions that each have the form:

  • Each expression has one input, either a Profile or a Relationship type. Each expression may contain an optional filter. Filters can be combined with logical operators.
  • An expression has an implied output Profile type, representing the the result set.
  • With the exception of the first expression, each expression can use as an input one of the following:
    • The same input profile or relationship type as the prior expression
    • The output profile type of the prior expression

For example, the following is a simple example of a query with the Salesperson profile type used explicitly as input and implicitly as output, and a single filter expression:

A segmentation clause is stored as a JSON string in the SegmentDefinition property of the Segment entity. Segmentation clauses can also be constructed interactively with the Segment Builder in Customer Insights.

Note: The segmentation query for a static segment is typically run only once, during it's initial construction (or explicit updating). In contrast, the segmentation query for a dynamic segment is reevaluated on a periodic basis, whose time interval is specified by the segmentRefreshRateIntervalMinutes property of the associated segment.

Query Elements

An expression is constructed of one or more of the following elements.

Query Element Description
PROFILE(profile-type-name) The starting expression for the query that selects all profiles of the specified profile type.
Limitation: queries using UNION, MINUS, INTERSECT or SEGMENT functions must consistently operate on the same profile type.
SEGMENT(segment-name) Selects all profiles belonging to a given segment
TRAVERSE(relationship-type-name, FILTER(condition)) Function to select the targets of the specified relationship. Optionally a filter condition can be supplied to the narrow the targets.
INTERACTION(interaction-type-name, link-type-name) Select all profiles linked to interaction. (Both parameters are required.) Similar to profile expression where queries can be run only on single interaction.
Limitation: INTERACTION can be used with FILTER and HAVING expressions but not with TRAVERSE.
HAVING(condition) Conditional expression that can be used on aggregated interactions
FILTER(condition) Applied to the current selection to exclude items not meeting the specified condition
NOT(condition) Applied to the current selection to exclude items that meet the specified condition
COUNT() Resolves to the number of items in the current selection.*
AVG(propertyName) Resolves to the average of the specified property values in the current selection.*
MIN(propertyName) Resolves to the smallest value of the specified property in the current selection.*
MAX(propertyName) Resolves to the largest value of the specified property in the current selection.*
SUM(propertyName) Resolves to the summation of the specified property values in the current selection.*
UNION Set operation that joins the resulting sets of two queries
EXCEPT Set operation that removes the results of the second query from results from the first query
INTERSECT Set operation that selects the items common to both queries
DATETIMEUTCNOW() Returns the current date and time (minutes precision) in UTC
DATEDIFF(datePart, date, date) Returns the count of the specified datepart boundaries crossed between the specified startdate and enddate. For complete documentation, see DATEDIFF (Transact-SQL).
DATEADD(datePart, number, date) Returns a specified date with the specified number interval added to a specified datepart of that date. For complete documentation, see DATEADD (Transact-SQL).

  * The aggregation functions COUNT, AVG, MIN, MAX, and SUM can be invoked inside of a HAVING expression, typically for a purpose of grouping a conditional expression.

Filtering Operations

The filtering operations supported in an expression depend upon the the data type of the target property as follows:

Property Type Supported Filtering Operations
Boolean equals
DateTime equals, greater-than, greater-than-or-equals, less-than, less-than-or-equals, not-equals
GUID equals, not-equals, one-of
Numeric (byte, decimal,
double, int, long, short)
equals, greater-than, greater-than-or-equals, less-than, less-than-or-equals, not-equals
String equals, not-equals, one-of


The following examples are based on a typical Dynamics 365 Customer Engagement schema.

Select all male contacts living in USA from the ages 15 to 25 and income over $50K:

      Sex == 'M' &&
      Country == 'USA' &&
      Age >= 15 && Age <= 25 &&
      Income > 50000

Select all contacts from leads where the lead date is after the end date of the campaign (of the lead’s marketing context):

      .FILTER(StartDate > '2017-03-30')

Select all company customers with at least one active purchase of a credit card where the expiration date is a month from today (useful to set up automated campaigns via rolling queries):

      NumberOfActivePurchasedItems > 0 && 
      Product == 'CreditCard' && 
      DATEDIFF(month, ExpirationDate, DATETIMEUTCNOW()) < 1)

Select all CEO contacts having accounts that bought "Product Xgenerator":

   .FILTER(Position == 'CEO')
      .FILTER(Name == 'Product Xgenerator')

Select all contacts that have clicked on any link in email "12345" in the last week:

INTERACTION(RedirectLinkClicked, RedirectLinkClicked_ContactLink)
   .FILTER(EmailId == '12345')

Select all customers whose birthday is within the next month:


Filter on all subscriptions that stem from this same email marketing campaign, but leaving out registrations from the first marketing event:

  .FILTER(Name == 'Campaign 1')
   .FILTER(Name == 'Event 1')

Create a compound segment that consists of segments AllFromSeattle and SubscribedToEvent: