Use Adobe Target and Web SDK for personalization

Adobe Experience Platform Web SDK can deliver and render personalized experiences managed in Adobe Target to the web channel. You can use a WYSIWYG editor, called the Visual Experience Composer (VEC), or a non-visual interface, the Form-based Experience Composer, to create, activate, and deliver your activities and personalization experiences.

The following features have been tested and are currently supported in Target:

• A/B Tests
• A4T Impression and conversion reporting
• Automated Personalization activities
• Experience Targeting activities
• Multivariate Tests (MVT)
• Recommendations activities
• Native Target impression and conversion reporting
• VEC Support

Web SDK system diagram

The following diagram helps you understand the workflow of Target and Web SDK edge decisioning.

Enabling Adobe Target

To enable Target, do the following:

1. Enable Target in your datastream with the appropriate client code.
2. Add the renderDecisions option to your events.

Then, optionally, you can also add the following options:

• decisionScopes: Retrieve specific activities (useful for activities created with the form-based composer) by adding this option to your events.
• Prehiding snippet: Hide only certain portions of the page.

Using the Adobe Target VEC

To use the VEC with a Web SDK implementation, install and activate either the Firefox or Chrome VEC Helper Extension.

For more information, see Visual Experience Composer helper extension in the Adobe Target guide.

Rendering personalized content

See Rendering personalization content for more information.

Audiences in XDM

When defining audiences for your Target activities that are delivered via the Web SDK, XDM must be defined and used. After you define XDM schemas, classes, and schema field groups, you can create a Target audience rule defined by XDM data for targeting. Within Target, XDM data displays in the Audience Builder as a custom parameter. The XDM is serialized using dot notation (for example, web.webPageDetails.name).

If you have Target activities with predefined audiences that use custom parameters or a user profile, they are not delivered correctly via the SDK. Instead of using custom parameters or the user profile, you must use XDM instead. However, there are out-of-the-box audience targeting fields supported via the Web SDK that do not require XDM. These fields are available in the Target UI that do not require XDM:

• Target Library
• Geo
• Network
• Operating System
• Site Pages
• Browser
• Traffic Sources
• Time Frame

For more information, see Categories for audiences in the Adobe Target guide.

Response tokens

Response tokens are used to send metadata to third parties like Google or Facebook. Response tokens are returned
in the meta field within propositions -> items. Here is a sample:

{
  "id": "AT:eyJhY3Rpdml0eUlkIjoiMTI2NzM2IiwiZXhwZXJpZW5jZUlkIjoiMCJ9",
  "scope": "__view__",
  "scopeDetails": ...,
  "renderAttempted": true,
  "items": [
    {
      "id": "0",
      "schema": "https://ns.adobe.com/personalization/dom-action",
      "meta": {
        "experience.id": "0",
        "activity.id": "126736",
        "offer.name": "Default Content",
        "offer.id": "0"
      }
    }
  ]
}

To collect the response tokens, you have to subscribe to alloy.sendEvent promise, iterate through propositions, and extract the details from items -> meta.

Every proposition has a renderAttempted boolean field indicating whether the proposition was rendered or not. See the sample below:

alloy("sendEvent",
  {
    "renderDecisions": true,
    "decisionScopes": [
      "hero-container"
    ]
  }).then(result => {
    const { propositions } = result;

    // filter rendered propositions
    const renderedPropositions = propositions.filter(proposition => proposition.renderAttempted === true);

    // collect the item metadata that represents the response tokens
    const collectMetaData = (items) => {
      return items.filter(item => item.meta !== undefined).map(item => item.meta);
    }

    const pageLoadResponseTokens = renderedPropositions
      .map(proposition => collectMetaData(proposition.items))
      .filter(e => e.length > 0)
      .flatMap(e => e);
  });

When automatic rendering is enabled, propositions array contains:

On Page-Load:

• Form-Based Composer based propositions with renderAttempted flag set to false
• Visual Experience Composer based propositions with renderAttempted flag set to true
• Visual Experience Composer based propositions for a Single Page Application view with renderAttempted flag set to true

On View - change (for cached views):

• Visual Experience Composer based propositions for a Single Page Application view with renderAttempted flag set to true

When automatic rendering is disabled, propositions array contains:

On Page-Load:

• Form-based Composer-based propositions with renderAttempted flag set to false
• Visual Experience Composer-based propositions with renderAttempted flag set to false
• Visual Experience Composer-based propositions for a Single Page Application view with renderAttempted flag set to false

On View - change (for cached views):

• Visual Experience Composer-based propositions for a Single Page Application view with renderAttempted flag set to false

Single profile update

The Web SDK lets you update the profile to the Target profile and to the Web SDK as an experience event.

To update a Target profile, ensure that the profile data is passed with the following:

• Under "data {"
• Under "__adobe.target"
• Prefix "profile."

Delay saving Profile or entity parameters until content has been displayed to the end user

To delay recording attributes in the profile until the content has been displayed, set data.adobe.target._save=false in your request.

For example, your website contains three decision scopes corresponding to three category links on the website (Men, Women and Kids) and you want to track the category the user eventually visited. Send these requests with the __save flag set to false to avoid persisting the category at the time the content is requested. After the content has been visualized, send the proper payload (including the eventToken and stateToken) for the corresponding attributes to be recorded.

The example below sends a trackEvent-style message, executes profile scripts, saves attributes, and immediately records the event.

alloy("sendEvent", {
    "renderDecisions": true,
    "xdm": { /* Experience Event XDM data */ },
    "data": {
        "__adobe": {
            "target": {
                " __save": true|false,
                //defaults to true if omitted
                "profile.gender": "female",
                "profile.age": 30,
                "entity.name": "T-shirt",
                "entity.id": "1234"
            }
        }
    }
})

Request recommendations

The following table lists Recommendations attributes and whether each one is supported via the Web SDK:

How to send Recommendations attributes to Adobe Target:

alloy("sendEvent", {
  "renderDecisions": true,
  "data": {
    "__adobe": {
      "target": {
        "entity.id": "123",
        "entity.genre": "Drama"
      }
    }
  }
});

Display mbox conversion metrics

The sample below shows how you can track display mbox conversions and send profile parameters to Adobe Target, without requiring to qualify for any content or activity.

alloy("sendEvent", {
    "xdm": {
        "_experience": {
            "decisioning": {
                "propositions": [{
                    "scope": "conversion-step-1" //example scope name
                }],
                "propositionEventType": {
                    "display": 1
                }
            }
        },
        "eventType": "decisioning.propositionDisplay"
    }
});

Debugging

mboxTrace and mboxDebug have been deprecated. Use a method of Web SDK debugging instead.

Terminology

Propositions: In Adobe Target, propositions correlate to the experience that is selected from an Activity.

Schema: The schema of a decision is the type of offer in Adobe Target.

Scope: The scope of the decision. In Adobe Target, the scope is the mBox. The global mBox is the __view__ scope.

XDM: The XDM is serialized into dot notation and then put into Adobe Target as mBox parameters.