Identity

The SDK surfaces a series of APIs allowing you to manage user-identity state. These client-side APIs work in tandem with the mParticle Identity HTTP API and your configured Identity “strategy.” These APIs are designed generically but identity management requirements vary by app - so it’s crucial that you use the APIs correctly per your requirements.

See the mParticle IDSync overview for a platform-agnostic overview of the key operations you can perform and read below for how the API is surfaced for the JavaScript.

Allowed Identity Types

IdentityType Description
customerid If you have an internal ID for your customer
email The user’s email address
facebook The user’s Facebook ID
facebookcustomaudienceid The user’s Facebook App User ID that can be retrieved through the Facebook SDK
google The user’s Google ID
twitter The user’s Twitter ID
microsoft The user’s Microsoft ID
yahoo The user’s Yahoo ID
other Any other identifier that can contribute to user identification

Creating a Request

The mParticle Identity API includes four key operations (identify, login, logout, and modify), and in the context of the Javascript SDK, all of these APIs accept a request object and an optional callback function. Populating this object correctly is crucial to managing the state of your users.

User Identities

An identity request object should be populated with a userIdentities object, which contains the identities that you would like to associate with the user. When you invoke any of the four key Identity APIs, the identities you supply will be associated with the resulting user.

var identityRequest = {
      userIdentities: {
          email:      'email@example.com',
          customerid: '123456'    
      }
   }

User Aliasing

The Identity API lets you transition the SDK and data from one user to a new or different user. This is implemented via the callback onUserAlias with arguments oldUser, and newUser, which both represent mParticle user objects. onUserAlias is only called if the user has changed. If while transitioning you’d like to copy any of the user attributes from the current user to the new user:

var identityRequest = {
      userIdentities: { email: 'email@example.com',
                        customerid: '123456'
      },
      onUserAlias: function(oldUser, newUser) {
          var oldUserAttributes = oldUser.getAllUserAttributes();
          for (var userAttrKey in oldUserAttributes){
              newUser.setUserAttibute(userAttrKey, oldUserAttributes[userAttrKey]);
          }
      }
}

The mParticle user object has a getCart method on it, which will retrieve the cart and can be used as follows:

var identityRequest = {
      userIdentities: { email: 'email@example.com',
                        customerid: '123456'
      },
      onUserAlias: function(oldUser, newUser) {
          newUser.getCart().add(oldUser.getCart().getCartProducts()); // getCartProducts() returns an array of products, but a single product can also be added to the `add` method
      }
}

Removing items is also possible with user.getCart().remove(product) or the cart can be completely cleared with user.getCart().clear(). Remove only allows for single products to be removed per method call, unlike add above, which takes both an array, or a single product.

Identify

The Identify API is treated specially in that it’s called automatically on SDK initialization. The SDK requires this call to succeed in order to establish an identity to associate with all data.

Some considerations to account for during SDK initialization:

  • If the user is already logged-in/registered on load of your site (from a previous session), or you otherwise have identifying information about the user that you’d like to supply, you should create a identity request and set it to the identifyRequest field of the mParticle.config object. See the SDK initialization example above.
  • If you do not provide an explicit identifyRequest during SDK initialization, the SDK will use browser local storage and/or cookies to generate a request for you based off of the most recent user, including the most recent user identities.
  • If this is a new user, and the Identify call fails, you should retry the request. See below for information on reacting to failed API requests.
  • You may want to set attributes directly on the user once identified. In order to do this, you should pass an identityCallback to the window.mParticle.config object as shown in the Advanced configuration options SDK snippet’s initialization. The callback is run once the identify returns. See below for an example:
window.mParticle.config = {
    rq: [],
    isDevelopmentMode: true,
    identifyRequest: {
        userIdentities: { email: 'email@example.com', customerid: '123456' }
    },
    identityCallback: myIdentityCallback
}

function myIdentityCallback(result) {
    if (result.httpCode === 200 || result.httpCode === -3) {
        // httpCode of 200 indicates is a successful identity call
        // httpCode of -3 indicates there was an active session, so no identity call is made, but the callback is still triggered
        result.getUser().setUserAttribute('age', 25);
    }
}

You can view additional httpCodes and details here in order to handle all errors/paths programmatically.

Login and Logout

Login and Logout should be invoked at the time of the user performing the matching or applicable actions on your site. These methods have identical signatures. They accept an identity request as above, as well as an optional callback function:

function onLoginResult(result){
  if (result.httpCode == 200) {
     //proceed with login
     //you can access the new/current mpid here
     var userId = result.body.mpid;
  } else if (result.httpCode == -1) {
    //browser has no coverage/HTTP request failed
  } else if (result.httpCode == 429) {
    //in rare circumstances, requests will be throttled and need to be retried
  }
}
mParticle.Identity.login(loginRequestUserObject, onLoginResult);
function onLogoutResult(result){
  if (result.httpCode == 200) {
     //proceed with logout
     //you can access the new/current mpid here
     var userId = result.body.mpid;
  } else if (result.httpCode == -1) {
    //browser has no coverage/HTTP request failed
  } else if (result.httpCode == 429) {
    //in rare circumstances, requests will be throttled and need to be retried
  }
}
mParticle.Identity.logout(logoutRequestUserObject, onLogoutResult);

Modify

Modify also has the identical signature, but note a crucial difference: modify actions will never result in a new user. Modify can only add, remove, or change the identities associated with an existing user. The mParticle SDK will compare the current user’s user identities with those that you supply within the identity request, in order to generate a delta and invoke the underlying Identity HTTP API.

In this example, the SDK will change the email of the current user, or add the email to the user’s profile if the user has no existing email on this device:

var modifyRequest = {
      userIdentities: { email: 'email@example.com' }
}
mParticle.Identity.modify(modifyRequest, onModifyResult);

In this example, the SDK will remove the email of the current user, or will be a no-op if the user has no email on this device:

var modifyRequest = {
      userIdentities: { email: null }
}
mParticle.Identity.modify(modifyRequest, onModifyResult);

Error Handling

The mParticle Identity API is intended to be central to your site’s state, and so is designed to be fast and highly-available. Similar to how your site may prevent users from logging in, logging out, or modifying their state without an internet connection - we intend you to treat these APIs as gating operations in order to maintain a consistent user state. The SDK will not retry API calls automatically, but provides a callback mechanism such that you can do so according to your business logic.

The SDK will always return the HTTP status and HTTP body of the underlying request:

function onIdentityResult(result) {
  if (result.httpCode == 200) {
    //proceed with the identity request
    //you can access the new/current mpid here
    var userId = result.body.mpid;
  } else if (result.httpCode == -1) {
      //browser has no coverage/HTTP request failed
  } else if (result.httpCode == -2) {
      //There is currently an active identity request in flight
  } else if (result.httpCode == -3) {
      // If there is an active session when the browser loads, the identify request is not run, but the callback still gets called
  } else if (result.httpCode == -4) {
      // there was a validation issue with your identityRequest object
  } else if (result.httpCode == 429) {
      //in rare circumstances, requests will be throttled and need to be retried
  }
}
mParticle.Identity.login(loginRequest, onIdentityResult);

mParticle User

Once the SDK has successfully called Identify for the first time, you will be able to access the current user via the mParticleUser object:

var currentUser = mParticle.Identity.getCurrentUser();
//query for the unique mParticle ID of this user
var mpid = currentUser.getMPID();
//Update user attributes associated with the user (there are several variations of this in addition to below)
currentUser.setUserAttribute("foo","bar");

Reserved Attributes

Below is a list of mParticle reserved user attributes that can be set:

  • $Age
  • $FirstName
  • $LastName
  • $Gender
  • $Mobile
  • $Address
  • $City
  • $State
  • $Zip
  • $Country

Attribute Keys

Always refer to your organization’s data plan when instrumenting user or event attributes. Each unique attribute key becomes a data point in the mParticle UI that can be filtered for each Output, used to drive the calculation of an Audience or become part of a Custom Mapping. This means that your choice of attribute keys can have a system-wide impact. For example, if you have a single attribute key per device that represents a unique User ID or a unique URL, and you have thousands of users, mParticle will see thousands of unique keys, even though you only create one per device.

Creating too many unique attribute keys can have several adverse effects:

  • The mParticle dashboard becomes overcrowded and it becomes harder for business users to manage individual data points.
  • Each individual data point can be switched on or off from the Data Filter. This filter information is downloaded by the SDK during your app’s initialization. The more unique data points you have, the larger the size of the download. If not controlled this can ultimately impact the performance of your app.
  • A high number of unique attribute keys makes it difficult for you to use mParticle features like Rules, Connection Settings and the Audience Builder to control your data flow. See examples.

You should avoid the following as attribute keys:

  • URLs
  • Dates
  • Dynamic strings
  • User IDs
  • Random IDs

Example

A gaming app has ten levels and you want to track which level each user has achieved.

Bad option: Create ten tags - reachedLevel1, reachedLevel2, reachedLevel3, etc.
Better option: Create a single attribute - reachedLevel = 4.

Capturing this data as a single attribute improves the performance of both your app and the mParticle dashboard by reducing the number of unique data points you need to manage. It’s also a much more useful data point. For example, you can easily create a single audience builder condition to target users within a range of levels reached.

Cross-Platform Attribute tracking

An mParticle Workspace can combine data from multiple platforms, for example it can show data from the same app running in iOS and Android. For this reason, you may wish to choose attribute names that you can keep consistent across all platforms for easier management. For example, if you call an attribute levelReached in iOS, LevelReached on Android, and level_reached on web, mParticle will treat these as three separate attributes.