Spec: Common Fields

In the Segment Spec all the API calls have a common structure, and a few common fields.

However, not all destinations accept all fields included in the Spec. Not sure which fields an destination accepts? Refer to the destination’s documentation page, or check out the open-source destination code on Github.


Every API call has the same core structure and fields. These fields describe user identity, timestamping and mechanical aides like API version.

Here’s an example of these common fields in raw JSON:

  "anonymousId": "507f191e810c19729de860ea",
  "context": {
    "active": true,
    "app": {
      "name": "InitechGlobal",
      "version": "545",
      "build": "",
      "namespace": "com.production.segment"
    "campaign": {
      "name": "TPS Innovation Newsletter",
      "source": "Newsletter",
      "medium": "email",
      "term": "tps reports",
      "content": "image link"
    "device": {
      "id": "B5372DB0-C21E-11E4-8DFC-AA07A5B093DB",
      "advertisingId": "7A3CBEA0-BDF5-11E4-8DFC-AA07A5B093DB",
      "adTrackingEnabled": true,
      "manufacturer": "Apple",
      "model": "iPhone7,2",
      "name": "maguro",
      "type": "ios",
      "token": "ff15bc0c20c4aa6cd50854ff165fd265c838e5405bfeb9571066395b8c9da449"
    "ip": "",
    "library": {
      "name": "analytics.js",
      "version": "2.11.1"
    "locale": "en-US",
    "location": {
      "city": "San Francisco",
      "country": "United States",
      "latitude": 40.2964197,
      "longitude": -76.9411617,
      "speed": 0
    "network": {
      "bluetooth": false,
      "carrier": "T-Mobile US",
      "cellular": true,
      "wifi": false
    "os": {
      "name": "iPhone OS",
      "version": "8.1.3"
    "page": {
      "path": "/academy/",
      "referrer": "",
      "search": "",
      "title": "Analytics Academy",
      "url": "https://segment.com/academy/"
    "referrer": {
      "id": "ABCD582CDEFFFF01919",
      "type": "dataxu"
    "screen": {
      "width": 320,
      "height": 568,
      "density": 2
    "groupId": "12345",
    "timezone": "Europe/Amsterdam",
    "userAgent": "Mozilla/5.0 (iPhone; CPU iPhone OS 9_1 like Mac OS X) AppleWebKit/601.1.46 (KHTML, like Gecko) Version/9.0 Mobile/13B143 Safari/601.1"
  "integrations": {
    "All": true,
    "Mixpanel": false,
    "Salesforce": false
  "messageId": "022bb90c-bbac-11e4-8dfc-aa07a5b093db",
  "receivedAt": "2015-12-10T04:08:31.909Z",
  "sentAt": "2015-12-10T04:08:31.581Z",
  "timestamp": "2015-12-10T04:08:31.905Z",
  "type": "track",
  "userId": "97980cfea0067",
  "version": 2

In more detail these common fields for every API call are:

anonymousIdoptional if userID is set insteadStringA pseudo-unique substitute for a User ID, for cases when you don’t have an absolutely unique identifier. A userId or an anonymousId is required.

See the Identities docs for more detail

contextoptionalObjectDictionary of extra information that provides useful context about a message, but is not directly related to the API call like ip address or locale

See the Context field docs for more detail

integrationsoptionalObjectDictionary of destinations to either enable or disable

See the Destinations field docs for more detail

messageIdimplicitStringAutomatically collected by Segment, a unique identifier for each message that lets you find an individual message across the API
receivedAtimplicitDateAutomatically set by Segment, the timestamp of when a message is received by Segment

It is an ISO-8601 date string. See the Timestamps fields docs for more detail

sentAtoptionalDateTimestamp of when a message is sent to Segment, used for clock skew correction

It is set automatically by the Segment tracking libraries. It is an ISO-8601 date string. See the Timestamps fields docs for more detail.

timestampoptionalDateTimestamp when the message itself took place, defaulted to the current time by the Segment Tracking API

It is an ISO-8601 date string

If the event just happened, leave it out and we’ll use the server’s time. If you’re importing data from the past, make sure you to provide a timestamp.See the Timestamps fields docs for more detail.

typeimplicitStringType of message, corresponding to the API method: 'identify', 'group', 'track', 'page', 'screen' or 'alias'
userIdoptional if anonymousID is set insteadStringUnique identifier for the user in your database

A userId or an anonymousId is required

See the Identities docs for more detail

versionimplicitNumberVersion of the Tracking API that received the message, automatically set by Segment

Beyond this common structure, each API call adds a few specialized top-level fields.


Context is a dictionary of extra information that provides useful context about a datapoint, for example the user’s ip address or locale. Context is a complete and explicit specification, so properties outside the spec are ignored. You should only use Context fields for their intended meaning.

activeBooleanWhether a user is active

This is usually used to flag an .identify() call to just update the traits but not “last seen.”

appObjectdictionary of information about the current application, containing name, version and build.

This is collected automatically from our mobile libraries when possible.

campaignObjectDictionary of information about the campaign that resulted in the API call, containing name, source, medium, term and content

This maps directly to the common UTM campaign parameters.

deviceObjectDictionary of information about the device, containing id, manufacturer, model, name, type and version
ipStringCurrent user’s IP address
libraryObjectDictionary of information about the library making the requests to the API, containing name and version
localeStringLocale string for the current user, for example en-US
locationObjectDictionary of information about the user’s current location, containing city, country, latitude, longitude, region and speed
networkObjectDictionary of information about the current network connection, containing bluetooth, carrier, cellular and wifi
osObjectDictionary of information about the operating system, containing name and version
pageObjectDictionary of information about the current page in the browser, containing hash, path, referrer, search, title and url

Automatically collected by Analytics.js.

referrerObjectDictionary of information about the way the user was referred to the website or app, containing type, name, url and link
screenObjectDictionary of information about the device’s screen, containing density, height and width
timezoneStringTimezones are sent as tzdata strings to add user timezone information which might be stripped from the timestamp

Ex: America/New_York

groupIdStringGroup / Account ID.

This is useful in B2B use cases where you need to attribute your non-group calls to a company or account. It is relied on by several Customer Success and CRM tools.

traitsObjectDictionary of traits of the current user

This is useful in cases where you need to track an event, but also associate information from a previous identify call. You should fill this object the same way you would fill traits in an identify call.

userAgentStringUser agent of the device making the request

Context Fields Automatically Collected

Below is a chart that shows you which context variables are populated automatically by our iOS, Android and analytics.js libraries.

Other libraries only collect context.library, any other context variables must be sent manually.

Context FieldAnalytics.jsAnalytics-iosAnalytics-android
  • IP Address is not collected by our libraries, but instead filled in by our servers when it receives a message for client side events only.
  • Our Android library collects screen.density with this method.


A dictionary of destination names that the message should be sent to. 'All' is a special key that applies when no key for a specific destination is found.

Integrations defaults to the following:

  All: true,
  Salesforce: false,

This is because Salesforce has strict limits on API calls, and we don’t want to run over your limits by accident.

Sending data to the rest of our destinations is opt-out so if you don’t specify the destination as false in this object, it will be sent to rest of the destinations that can accept it.


Every API call has four timestamps, originalTimestamp, timestamp, sentAt and receivedAt. They’re used for very different purposes.

All timestamps are ISO-8601 date strings.

Timestamp Overview


Time on the client device when call was invoked


timestamp value manually passed in through server-side libraries

Used by Segment to calculate timestamp

Note: originalTimestamp is not useful for analysis since it’s not always trustworthy as it can be easily adjusted and affected by clock skew


Time on client device when call was sent


sentAt value manually passed in

Used by Segment to calculate timestamp

Note: sentAt is not useful for analysis since it’s not always trustworthy as it can be easily adjusted and affected by clock skew

receivedAttime on Segment server clock when call was received

Used by Segment to calculate timestamp

Used as sort key in Warehouses

Note: For max query speed, receivedAt is the recommended timestamp for analysis when chronology does not matter as chronology is not ensured


Calculated by Segment to correct client-device clock skew using the following formula:

receivedAt - (sentAt - originalTimestamp)

Used by Segment to send to downstream destinations

Used for historical replays

Note: Recommended timestamp for analysis when chronology does matter


The originalTimestamp tells you when call was invoked on the client device or the value of timestamp that you manually passed in.

Note: The originalTimestamp timestamp is not useful for any analysis since it’s not always trustworthy as it can be easily adjusted and affected by clock skew.


The sentAt timestamp specifies the clock time for the client’s device when the network request was made to the Segment API. For libraries and systems that send batched requests, there can be a long gap between a datapoint’s timestamp and sentAt. Combined with receivedAt, we can use sentAt to correct the original timestamp in situations where a user’s device clock cannot be trusted (mobile phones and browsers). The sentAt and receivedAt timestamps are assumed to occur at the same time (maximum a few hundred milliseconds), and therefore the difference is the user’s device clock skew, which can be applied back to correct the timestamp.

Note: The sentAt timestamp is not useful for any analysis since it’s tainted by user’s clock skew.


The receivedAt timestamp is added to incoming messages as soon as they hit our API. It’s used in combination with sentAt to correct clock skew, and also to aid with debugging libraries and systems that deliver events in batches.

The receivedAt timestamp is most important as the sort key in our Warehouses product. Use this for max query speed when retrieving data from your Warehouse!

Note: Chronological order of events is not ensured with receivedAt.


The timestamp timestamp specifies when the datapoint occurred, corrected for client-device clock skew. This is the timestamp that is passed to downstream destinations and used for historical replays. It is important to use this timestamp for importing historical data to the API.

The timestamp field is settable from our server-side libs or if passing info directly to the HTTP endpoints.

Questions? Need help? Contact us!
Can we improve this doc?
Email us: docs-feedback@segment.com!