Marketing Campaigns V2 Segmentation Query Reference

This document is a reference for the V2 Segmentation API. The V2 Segmentation API allows you to create and manage segments using a subset of SQL that includes statements and operations necessary to define any segment. In addition, this version of the Segmentation API exposes the contact_data and event_data tables to enable SQL-compatible queries.

General Query Format

SELECT contact_id, updated_at
FROM <table_reference> [ AS <table_alias> ]
[ [ INNER ] JOIN <table_reference> ON contact_id ]
[ WHERE <predicate> ]

In the previous example, the contact_id and updated_at columns are selected in the query. These two columns must be selected, and aggregate function MAX and its alias can be selected part of Last Event queries. Other than those columns, no additional columns can be selected. When joining two tables, they must be joined on the contact_id column only because it is the foreign key used to connect the contact_data and event_data tables.

All Available Query Keywords

• SELECT
• FROM
• JOIN (defaults to INNER JOIN)
• INNER JOIN
• ON
• WHERE

Available Properties

contact_data(required)

event_data(optional)

Event data that can be optionally be used to enhanced segments

Possible Event Types

• click - Whenever a recipient clicks one of the Click Tracked links in your email. In the Email History, SendGrid displays the date, time, and the URL for the link that was clicked.
• blocked - When your IP address has been blocked by an ISP or messaging organization. Blocks are less severe than bounces and do not result in permanent suppressions: subsequent sends to blocked email addresses are not automatically suppressed.
• bounce - The receiving server could not or would not accept the message. If a recipient has previously unsubscribed from your emails, your attempt to send to them is bounced.
• deferred - The recipient mail server asked SendGrid to stop sending emails so fast.
• delivered - The accepted response generated by the recipients' mail server.
• dropped - Twilio SendGrid will drop an email when the contact on that email is in one of your suppression groups, the recipient email previously bounced, or that recipient has marked your email as spam.
• group_resubscribe - When a recipient resubscribes themselves to a suppression group.
• group_unsubscribe - Whenever a recipient unsubscribes from a suppression group.
• open - The response generated by a recipient opening an email.
• processed - Requests from your website, application, or mail client via SMTP Relay or the API that SendGrid processed.
• spamreport - Whenever a recipient marks your email as spam and their mail server tells us about it.
• unsubscribe - Whenever a recipient unsubscribes from your emails.

Event_Data JSON Struct Example

Supported Functions and examples

current_timestamp

Example query

current_timestamp

The following query will return all contacts that were created before this run of the segment:

SELECT contact_id, updated_at
FROM contact_data
WHERE created_at < CURRENT_TIMESTAMP

timestampadd(interval, count, timestamp)

timestampadd allows you to specify a time interval, a number of those intervals as a count, and a timestamp that you can use to compare Date fields against.

Parameters

Example Query

timestampadd(interval, count, timestamp)

The following query will return all contacts that were created before one year prior to 2020-10-15 6PM GMT:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE created_at > timestampadd(day, 1, '2020-10-15T18:00:12Z')

array_contains(field_name, array)

array_contains checks for at least one of the values provided in an array. It does not hard check for the presence of all the values or an exact match of the values given. It also doesn't support pattern-matches like '%gmail.com%' as array values. array_contains will instead check for the exact string, including pattern match characters. For example, when searching '%gmail.com%' the % characters are considered part of the string literal.

Parameters

Example Query

array_contains(field_name, array)

The following query will return all contacts with email domains equal to gmail.com or yahoo.com:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE array_contains(email_domains, ['gmail.com', 'yahoo.com'])

contains_word(word_value, field_name)

contains_word is a custom function provided by the Marketing Campaigns V2 Segments API. contains_word accepts two arguments: a word_value that will be searched for in the field_name values stored for each contact.

The field_name you specify must be of type Text, and the value stored in field_name must be a comma-separated string. For example, if the function evaluates the word_value alice against a first_name field where the value is bob,alice,sue, it will find alice and return the contact. However, if the first_name value is stored as bobalicesue, bob.alice.sue, or any other non comma separated string, alice will not be found, and the contact will not be returned. This is important when thinking about how to enter the string values associated with your contacts. The check is case sensitive.

• All special characters and spaces within the comma separated string will be included when looking for a match. For Example, a field with a value of bob,(alice),sue and a match value of (alice) will return a contact
• Single quotes in a field value must be escaped in the match value using double single quotes. For Example, a field value of bo'b,alice will require a match value of bo''b.

Parameters

Example Query

The following query will return all contacts where first_name contains the word bob :

SELECT contact_id, updated_at 
FROM contact_data 
WHERE contains_word('bob', first_name)

Sample Queries for Different Kinds of Segments

Segmentation Examples From All Contacts

All contacts:

SELECT contact_id, updated_at 
FROM contact_data

All contacts with first_name ‘Dave’:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE first_name = 'Dave'

All contacts where state_province_region is ‘Colorado’:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE state_province_region = 'CO'

All contacts with primary email with the substring ‘gmail.com’:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE email like '%gmail.com%'

All contacts with a text type custom field my_text_custom_field value ‘abc’:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE my_text_custom_field = 'abc'

All contacts with primary email with ‘gmail.com’ as domain name, and a text type custom field my_custom_field value ‘abc’:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE email like '%gmail.com' and my_custom_field = 'abc'

All contacts with a number type custom field my_number_custom_field value 12:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE my_number_custom_field = 12

All contacts with a date type custom field my_date_custom_field value `2021-01-01T12:46:24Z’:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE my_date_custom_field = '2021-01-01T12:46:24Z'

All contacts where alternate email is equal to ‘alternate@gmail.com’:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE array_contains(alternate_emails, ['alternate@gmail.com'])

All contacts where alternate email is equal to ‘alternate@gmail.com’ or ‘alternate2@gmail.com’:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE array_contains(alternate_emails, ['alternate@gmail.com','alternate2@gmail.com'])

All contacts present in a specific list “list_id”:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE array_contains(list_ids, ['02bc41ca-07f9-11eb-9c1f-422f9e0f60d1'])

All contacts present in either of the list_ids:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE array_contains(list_ids, ['02bc41ca-07f9-11eb-9c1f-422f9e0f60d1', '02bc41ca-07f9-11eb-9c1f-422f9e0f62e4', '042a8e48-6e31-11eb-a8bc-7656c249c550'])

All contacts with specific email domain(s) gmail.com:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE array_contains(email_domains, ['gmail.com’])

All contacts where created_at is after 2021-01-01 12 PM GMT:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE created_at > '2021-01-01T12:00:00Z'

All contacts where created_at is equal to 2021-01-01 12 PM GMT:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE created_at = '2021-01-01T12:00:12Z'

All contacts with external id that starts with ‘123’:

SELECT contact_id, updated_at
FROM contact_data
WHERE external_id like '123%'

NULL Values

When a contact's reserved fields are not set during creation, they default to a NULL value. Not setting a value for custom fields when creating a contact will save the contact without those custom fields. A segment query using these fields does not return contacts having NULL as a value. Additional conditions may be used to specify the field is null so that contacts having a null value will be included in the segment.

All contacts where first_name does not equal Dave, also including contacts without a value for first_name:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE first_name != ‘Dave’ OR first_name is null

All contacts where my_custom_field does not contain abc, also including contacts without a value for my_custom_field:

SELECT contact_id, updated_at 
FROM contact_data 
WHERE my_custom_field like ‘%abc%’ OR my_custom_field is null

JOIN Types

Only JOIN and INNER JOIN are allowed and INNER JOIN will be internally converted to JOIN. LEFT JOIN and RIGHT JOIN are restricted for performance, and the same functionality can be achieved using JOIN. The examples below show how LEFT JOIN / RIGHT JOIN can be replaced with JOIN and UNION.

The following queries return all contacts that have an event “delivered” for a particular Single Send or whose state is Colorado. While both queries return the same set of contacts, using JOIN and UNION is more performant than using LEFT JOIN.

LEFT JOIN (not supported)

SELECT c.contact_id, c.updated_at 
FROM contact_data c
LEFT JOIN event_data e 
ON c.contact_id = e.contact_id
WHERE e.event_source = 'mail'
AND e.event_type = 'delivered' 
AND e.DATA:payload.unique_args.singlesend_id = '042a8e48-6e31-11eb-a8bc-7656c249c550'
OR c.state_province_region = 'CO'

JOIN and UNION

UNION can be used instead of "OR" to return contacts based on distinct criteria that requires use of a JOIN.

For example, a query for contacts that were 'delivered' a Single Send OR the contacts whose state_province_region is in "CO" can be written as follows:

SELECT c.contact_id, c.updated_at 
FROM contact_data c
JOIN event_data e 
ON c.contact_id = e.contact_id
WHERE e.event_source = 'mail'
AND e.event_type = 'delivered'
AND e.DATA:payload.unique_args.singlesend_id = '042a8e48-6e31-11eb-a8bc-7656c249c550'
UNION 
SELECT contact_id, updated_at 
FROM contact_data 
WHERE c.state_province_region = 'CO'

Engagement and Multi-engagement Examples

All contacts that have at least one event ( i.e. an attempt was made to send them an email):

SELECT c.contact_id, c.updated_at 
FROM contact_data c 
JOIN event_data e ON c.contact_id = e.contact_id

All contacts that have opened emails from two different single sends:

SELECT e2.contact_id, c2.updated_at
FROM event_data e2
INNER JOIN (
 SELECT c1.contact_id, c1.updated_at
 FROM contact_data AS c1
 INNER JOIN event_data e1 ON c1.contact_id = e1.contact_id
 WHERE e1.event_source = 'mail'
   AND e1.event_type = 'open'
   AND e1.DATA:payload.unique_args.singlesend_id = '02bc41ca-07f9-11eb-9c1f-422f9e0f60d1'
) AS c2 ON c2.contact_id = e2.contact_id
WHERE e2.event_source = 'mail' 
 AND e2.event_type = 'open'
 AND e2.DATA:payload.unique_args.singlesend_id = '00163f67-7211-4363-ab4e-12dd6f313b3a'

All contacts that have opened ANY single sends within 3 days:

SELECT c.contact_id, c.updated_at
FROM contact_data as c
JOIN event_data as e on c.contact_id = e.contact_id
WHERE e.event_source = 'mail'
    AND e.event_type = 'open'
    AND e.DATA:payload.unique_args.singlesend_id is not null
    AND e.timestamp >= timestampadd(day,  -3,  current_timestamp())

All contacts that have both of the two events - click and open:

SELECT e2.contact_id, c2.updated_at
FROM event_data e2
INNER JOIN (
 SELECT c1.contact_id, c1.updated_at
 FROM contact_data AS c1
 INNER JOIN event_data e1 ON c1.contact_id = e1.contact_id
 WHERE e1.event_source = 'mail' AND e1.event_type = 'open'
) AS c2 ON c2.contact_id = e2.contact_id
WHERE e2.event_source = 'mail' AND e2.event_type = 'click'

All contacts that have both of the two events - bounce and deferred for a particular automation:

SELECT e2.contact_id, c2.updated_at
FROM event_data e2
INNER JOIN (
 SELECT c1.contact_id, c1.updated_at
 FROM contact_data AS c1
 INNER JOIN event_data e1 ON c1.contact_id = e1.contact_id
 WHERE e1.event_source = 'mail' 
  AND e1.event_type = 'bounce'
  AND e1.DATA:payload.unique_args.mc_auto_id = '02bc41ca-07f9-11eb-9c1f-422f9e0f60d1'
) AS c2 ON c2.contact_id = e2.contact_id
WHERE e2.event_source = 'mail' 
 AND e2.event_type = 'deferred'
 AND e2.DATA:payload.unique_args.mc_auto_id = '02bc41ca-07f9-11eb-9c1f-422f9e0f60d1'

Not Event Examples

All contacts that have no event data (i.e. all contacts with where there has been no attempt to send an email to):

SELECT c.contact_id, c.updated_at
FROM contact_data c
WHERE c.contact_id NOT IN (
 SELECT e.contact_id
 FROM event_data e
)

All contacts that have not opened any mail in the last two months worth of seconds from the time when the segment is run:

SELECT c.contact_id, c.updated_at
FROM contact_data c
WHERE c.contact_id NOT IN (
 SELECT e.contact_id
 FROM event_data e
 WHERE e.event_source = 'mail' 
   AND e.event_type = 'open'
   AND e.timestamp < timestampadd(MONTH, -2, CURRENT_TIMESTAMP)
)

All contacts that have been 'delivered' the ANY single sends but have not 'open' ANY single sends within 1 month:

SELECT c.contact_id, c.updated_at
FROM contact_data c
JOIN event_data AS e1 ON c.contact_id = e1.contact_id
 WHERE e1.event_source = 'mail' 
 AND e1.event_type = 'delivered'
 AND e1.DATA:payload.unique_args.singlesend_id is not null
 AND c.contact_id NOT IN (
   SELECT e2.contact_id
   FROM event_data AS e2
   WHERE e2.event_source = 'mail' 
     AND e2.event_type = 'open'
     AND e2.DATA:payload.unique_args.singlesend_id is not null
     AND e2.timestamp >= timestampadd(month, -1, current_timestamp()))
 )

All contacts that have been 'delivered' the second email from automation but have not 'open' the email (mc_auto_id = 'fa3e0f6e-d397-11eb-87ce-22ffc6ed50f5' mc_auto_step_id = 'fc9ead9f-d397-11eb-92fe-02aa17a6534e'):

SELECT c.contact_id, c.updated_at
FROM contact_data c
JOIN event_data AS e1 ON c.contact_id = e1.contact_id
 WHERE e1.event_source = 'mail' 
 AND e1.event_type = 'delivered'
 AND e1.DATA:payload.unique_args.mc_auto_id = 'fa3e0f6e-d397-11eb-87ce-22ffc6ed50f5'
 AND e1.DATA:payload.unique_args.mc_auto_step_id = 'fc9ead9f-d397-11eb-92fe-02aa17a6534e'
 AND c.contact_id NOT IN (
   SELECT e2.contact_id
   FROM event_data AS e2
   WHERE e2.event_source = 'mail' 
     AND e2.event_type = 'open'
     AND e2.DATA:payload.unique_args.mc_auto_id = 'fa3e0f6e-d397-11eb-87ce-22ffc6ed50f5'
     AND e2.DATA:payload.unique_args.mc_auto_step_id = 'fc9ead9f-d397-11eb-92fe-02aa17a6534e'
 )

Email Activity - Last Event Examples

There are three email activities that support last events: “Last Clicked”, “Last Opened”, and “Last Emailed”. The last event queries are helpful to see the contacts whose last mail activity is in a certain period. The following queries are used to segment contacts based on their last mail activity.

All contacts that have not opened contacts since January of 2021 (last opened activity is before 2021):

SELECT c1.contact_id, c1.updated_at 
  FROM contact_data as c1 
  JOIN (
    SELECT e1.contact_id, e1.max_timestamp 
    FROM (
      SELECT e.contact_id, MAX(e.timestamp) max_timestamp 
      FROM event_data e 
      WHERE e.event_source = 'mail' and e.event_type = 'open' 
      GROUP BY e.contact_id
    ) as e1
    WHERE e1.max_timestamp < '2021-01-01T00:00:00.000Z'
  ) as e2 on c1.contact_id = e2.contact_id

All contacts that last clicked an email within 7 days (last clicked activity is within 7 days):

SELECT c1.contact_id, c1.updated_at 
  FROM contact_data as c1 
  JOIN (
    SELECT e1.contact_id, e1.max_timestamp 
    FROM (
      SELECT e.contact_id, MAX(e.timestamp) max_timestamp 
      FROM event_data e 
      WHERE e.event_source = 'mail' and e.event_type = 'click' 
      GROUP BY e.contact_id
    ) as e1
    WHERE e1.max_timestamp >= timestampadd(day, -7, current_timestamp())
  ) as e2 on c1.contact_id = e2.contact_id

All contacts that were last emailed in the past three months and last opened in the previous month (last emailed is within 3 months and last opened is within 1 month):

SELECT c1.contact_id, c1.updated_at 
FROM (
    SELECT c.contact_id, c.updated_at 
    FROM contact_data as c 
    JOIN (
        SELECT e1.contact_id, e1.max_timestamp 
        FROM (
            SELECT e.contact_id, MAX(e.timestamp) as max_timestamp 
            FROM event_data e 
            WHERE e.event_source = 'mail' 
             AND e.event_type = 'delivered' group by e.contact_id
        ) as e1 
    WHERE e1.max_timestamp >= timestampadd(month, -3, current_timestamp())
    ) as e2 on c.contact_id = e2.contact_id
) as c1 
JOIN (
    SELECT c.contact_id, c.updated_at 
    FROM contact_data as c 
    JOIN (
         SELECT e1.contact_id, e1.max_timestamp 
         FROM (
              SELECT e.contact_id, MAX(e.timestamp) as max_timestamp 
              FROM event_data e 
              WHERE e.event_source = 'mail' 
               AND e.event_type = 'open' group by e.contact_id
         ) as e1 
    WHERE e1.max_timestamp >= timestampadd(month, -1, current_timestamp())
    ) as e2 on c.contact_id = e2.contact_id
) as c11 on c1.contact_id = c11.contact_id

JOINing More Than 2 Data Sets

Multiple data sets are allowed to be joined together if done in a specific manner even though SQL supports both JOINs expressed linearly and JOINs expressed using subqueries. Each data set’s criteria must be listed alongside the corresponding table reference as it makes for better readability by having the queries nested explicitly by using subqueries. Following is an example of a segment query for both types. Here, the query defines a segment of all contacts that have the event of type 'processed' from the list of contacts which have the event of type 'delivered'.

Concatenated JOINs (not supported)

SELECT c1.contact_id, c1.updated_at FROM event_data AS e2
JOIN contact_data AS c1 ON e2.contact_id = c1.contact_id
JOIN event_data as e1 ON e1.contact_id = e2.contact_id
WHERE e1.event_source = 'mail' AND e1.event_type = 'delivered' AND e2.event_source = 'mail' AND e2.event_type = 'processed'

This logic, however, can easily be represented using JOINs with subqueries in a more readable way.

JOINs with subqueries

SELECT e2.contact_id, c2.updated_at
FROM event_data AS e2 JOIN (
    SELECT c1.contact_id, c1.updated_at
    FROM contact_data AS c1 JOIN event_data AS e1
    ON c1.contact_id = e1.contact_id
    WHERE e1.event_source = 'mail' AND e1.event_type = 'delivered') AS c2
ON c2.contact_id = e2.contact_id
WHERE  e2.event_source = 'mail' AND e2.event_type = 'processed'

Curl Usage

When creating or changing a segment query using curl command, escape any single quotes present in any parameter:

SELECT contact_id, updated_at
FROM contact_data 
WHERE first_name = '\''Dave'\''