Optimizely

Integrating Optimizely's experiment tools with Amplitude Analytics lets you share user data in real-time and helps non-technical product and marketing team members to do full-circle tracking, analysis, and engagement across web and mobile apps.

This integration is available to all customers who have access to Optimizely's Custom Analytics Integration feature.

Before you begin

Be sure you have the most recent Amplitude Browser SDK implemented. Also, you must place the Amplitude snippet before the Optimizely snippet.

Set up and use the integration

To integrate Optimizely with Amplitude, follow these steps:

1. Log in to your Optimizely account, select a project, and navigate to Settings > Integrations. The URL should adhere to this format: https://app.optimizely.com/v2/projects/[PROJECT_ID]/settings/integrations

2. From the Create Analytics Integration.. dropdown, select Using JSON.

3. Insert the following code:

    1{
    2  "plugin_type":"analytics_integration",
    3  "name":"Amplitude Analytics Integration",
    4  "form_schema":[
    5    {
    6      "default_value":"[Optimizely Experiment]",
    7      "field_type":"text",
    8      "name":"property_prefix",
    9      "label":"User Property Prefix: A user property for your experiment will be sent in an identify call. The user property will be prefixed with this prefix.",
   10      "options":null
   11    },
   12    {
   13      "default_value":"n",
   14      "field_type":"dropdown",
   15      "name":"send_event",
   16      "label":"Send Event: Optionally send an event with campaign, experiment, and variation info",
   17      "options":{
   18        "choices":[
   19          {
   20            "value":"n",
   21            "label":"No (not needed)"
   22          },
   23          {
   24            "value":"y",
   25            "label":"Yes"
   26          }
   27        ]
   28      }
   29    },
   30    {
   31      "default_value":"User in Experiment",
   32      "field_type":"text",
   33      "name":"event_name",
   34      "label":"Event Name (optional)",
   35      "options":null
   36    },
   37    {
   38      "default_value":"",
   39      "field_type":"text",
   40      "name":"instance_name",
   41      "label":"Instance Name (optional): The instance name you are using in your amplitude instrumentation eg: amplitude.getInstance('my instance name'). Usually this is blank.",
   42      "options":null
   43    }
   44  ],
   45  "description":"Send an identify call to amplitude identifying the experiment variation the user is seeing. Also, optionally send an event to amplitude that the user is in an experiment.\n\nSettings:",
   46  "options":{
   47    "track_layer_decision":"// VERSION 0.1.0\n// Last Updated: July 9th 2019\n\nvar dataSent = false;\nvar MAX_ATTEMPTS = 9;\nvar RETRY_DELAY_MS = 1000;\n\nfunction getCampaignInfo() {\n return window.optimizely\n .get(\"state\")\n .getDecisionObject({ campaignId: campaignId });\n}\n\nfunction logEvent() {\n var campaignInfo = getCampaignInfo();\n\n if (campaignInfo) {\n var eventProperties = {\n \"[Optimizely Campaign]\": campaignInfo.campaign,\n \"[Optimizely Experiment]\": campaignInfo.experiment,\n \"[Optimizely Variation]\": campaignInfo.variation,\n \"[Optimizely Holdback]\": campaignInfo.holdback\n };\n amplitude.getInstance(extension.instance_name).logEvent(extension.event_name, eventProperties);\n }\n}\n\nfunction identifyCall() {\n var campaignInfo = getCampaignInfo();\n\n if (campaignInfo) {\n var identify = new amplitude.Identify().set(\n extension.property_prefix + \" \" + campaignInfo.experiment,\n campaignInfo.variation\n );\n amplitude.getInstance(extension.instance_name).identify(identify);\n }\n}\n\nfunction sendData() {\n if (!dataSent) {\n identifyCall();\n if (extension.send_event === \"y\") {\n logEvent();\n }\n }\n dataSent = true;\n}\n\nfunction sendToAmplitude(call) {\n if (call >= MAX_ATTEMPTS) {\n return;\n }\n \n var instanceKey = extension.instance_name || \"$default_instance\";\n\n if (window.amplitude && window.amplitude.getInstance) {\n var instance = window.amplitude.getInstance(extension.instance_name);\n \n if (instance._isInitialized) {\n return sendData();\n } else if (instance.onInit) {\n instance.onInit(function() {\n sendData();\n });\n return;\n }\n }\n \n return setTimeout(function() {\n sendToAmplitude(call + 1);\n }, RETRY_DELAY_MS);\n}\nsendToAmplitude(0);"
   48  }
   49}

4. Turn the integration from Optimizely's Project Settings page. It's listed as Amplitude Analytics Integration.

5. Navigate to an experiment that's generating data that you want in Amplitude. In the Optimizely experiment settings panel, go to Integrations, and enable Tracked for the Amplitude Analytics Integration.
   You can also take this opportunity to adjust the following settings:

   1. User Property Prefix: Optimizely sends a user property to Amplitude to identify users who have been bucketed into the Optimizely experiment. This prefix setting makes it easier to identify these properties in Amplitude.
   2. Send Event [Optional]: When enabled, sends an event every time an experiment is fired. Configure to either yes or no.
   3. Event Name [Optional]: If you've selected to send an event, use this setting to specify it. Example: "Viewed Experiment"
   4. Instance Name [Optional]: See the Amplitude JavaScript SDK docs for more information about instance names.

6. Save your settings.

7. When the Optimizely experiment is running and collecting data, the user properties for this experiment appear in Amplitude.