Collecting data with Trackers and Webhooks

  1. Home
  2. Docs
  3. Collecting data with Trackers and Webhooks
  4. Trackers – collecting data from your own applications
  5. Javascript Tracker
  6. General Parameters
  7. Initializing a Tracker

Initializing a Tracker

Initialising a tracker

Tracker initialization is indicated with the "newTracker" string and takes three arguments:

  1. The tracker namespace
  2. The collector endpoint
  3. An optional argmap containing other settings

Here is a simple example of how to initialise a tracker:

snowplow_name_here("newTracker", "sp", "<<collector_url_here>>", { appId: "cfe23a", platform: "web", contexts: { webPage: true, performanceTiming: true } });

The tracker will be named “sp” and will send events to the a collector url you specify by replacing <<collector_url_here>>. The final argument is called the argmap. Here it is just used to set the app ID and platform for the tracker and a couple of common contexts for each event. Each event the tracker sends will have an app ID field set to “cfe23a” and a platform field set to “web”.

Here is a longer example in which every tracker configuration parameter is set:

snowplow_name_here("newTracker", "sp", "<<collector_url_here>>", { appId: "cfe23a", platform: "web" cookieDomain: null, discoverRootDomain: true, cookieName: "_sp_", cookieSameSite: "None", cookieSecure: true, encodeBase64: true, respectDoNotTrack: false, pageUnloadTimer: 500, forceSecureTracker: false, eventMethod: "post", bufferSize: 1, maxPostBytes: 40000, crossDomainLinker: function (linkElement) { return (linkElement.href === "http://acme.de" || linkElement.id === "crossDomainLink"); }, cookieLifetime: 63072000, stateStorageStrategy: "cookieAndLocalStorage", maxLocalStorageQueueSize: 1000, resetActivityTrackingOnPageView: true, contexts: { webPage: true, performanceTiming: true, gaCookies: true, geolocation: false } });

We will now go through the various argmap parameters. Note that these are all optional. In fact, you aren’t required to provide any argmap at all.

Setting the application ID

Set the application ID using the appId field of the argmap. This will be attached to every event the tracker fires. You can set different application IDs on different parts of your site. You can then distinguish events that occur on different applications by grouping results based on application_id.

Setting the platform

Set the application platform using the platform field of the argmap. This will be attached to every event the tracker fires. Its default value is “web”. For a list of supported platforms, please see the Snowplow Tracker Protocol.

Configuring the cookie domain

If your website spans multiple subdomains e.g.

You will want to track user behaviour across all those subdomains, rather than within each individually. As a result, it is important that the domain for your first party cookies is set to ‘.mysite.com’ rather than ‘www.mysite.com’. By doing so, any values that are stored on the cookie on one of subdomain will be accessible on all the others.

It is recommended that you enable automatic discovery and setting of the root domain.

Otherwise, set the cookie domain for the tracker instance using the cookieDomain field of the argmap. If this field is not set, the cookies will not be given a domain.

WARNINGChanging the cookie domain will reset all existing cookies. As a result, it might be a major one-time disruption to data analytics because all visitors to the website will receive a new domain_userid.

Configuring the cookie name

Set the cookie name for the tracker instance using the cookieName field of the argmap. The default is “sp“. Snowplow uses two cookies, a domain cookie and a session cookie. In the default case, their names are “_sp_id” and “_sp_ses” respectively. If you are upgrading from an earlier version of Snowplow, you should use the default cookie name so that the cookies set by the earlier version are still remembered. Otherwise you should provide a new name to prevent clashes with other Snowplow users on the same page.

Once set, you can retrieve a cookie name thanks to the getCookieName(basename) method where basename is id or ses for the domain and session cookie respectively. As an example, you can retrieve the complete name of the domain cookie with getCookieName('id').

Configuring the cookie samesite attribute

Set the cookie samesite attribute for the tracker instance using the cookieSameSite field of the argmap. The default is “None”. Valid values are “Strict”, “Lax”, “None” or “null”. “null” will not set the SameSite attribute.

It is recommended to set either “None” or “Lax”. You must use “None” if using the tracker in a third party iframe. “Lax” is good in all other cases and must be used if not setting Secure to true.

Configuring the cookie secure attribute

Set the cookie secure attribute for the tracker instance using the cookieSecure field of the argmap. The default is “true”. Valid values are “true” or “false”.

It is recommended to set this to “true”. This must be set to “false” if using the tracker on non-secure HTTP.

Configuring base 64 encoding

By default, self-describing events and custom contexts are encoded into Base64 to ensure that no data is lost or corrupted. You can turn encoding on or off using the encodeBase64 field of the argmap.

Respecting Do Not Track

Most browsers have a Do Not Track option which allows users to express a preference not to be tracked. You can respect that preference by setting the respectDoNotTrack field of the argmap to true. This prevents cookies from being sent and events from being fired.

Opt-out cookie

It is possible to set an opt-out cookie in order not to track anything similarly to Do Not Track through window.snowplow_name_here('setOptOutCookie', 'opt-out'); where ‘opt-out’ is the name of your opt-out cookie. If this cookie is set, cookies won’t be stored and events won’t be fired.

User fingerprinting

User Fingerprinting has been removed as of 2.13.0.

Setting the user fingerprint seed

User Fingerprinting has been removed as of 2.13.0.

Setting the page unload pause

Whenever the Snowplow Javascript Tracker fires an event, it automatically starts a 500 millisecond timer running. If the user clicks on a link or refreshes the page during this period (or, more likely, if the event was triggered by the user clicking a link), the page will wait until either the event is sent or the timer is finished before unloading. 500 milliseconds is usually enough to ensure the event has time to be sent.

You can change the pause length (in milliseconds) using the pageUnloadTimer of the argmap. The above example completely eliminates the pause. This does make it unlikely that events triggered by link clicks will be sent.

See also How the Tracker uses localStorage for an explanation of how the tracker can later recover and send unsent events.

Setting the event request protocol

Normally the protocol (http or https) used by the Tracker to send events to a collector is the same as the protocol of the current page. You can force it to use https by setting the forceSecureTracker field of the argmap to true.

Setting an unsecure event request protocol

Normally the protocol (http or https) used by the Tracker to send events to a collector is the same as the protocol of the current page. You can force it to use http by setting the forceUnsecureTracker field of the argmap to true. If forceSecureTracker is activated this argument is ignored.

NOTE: This argument should only be used for testing purposes as it creates security vulnerabilities.

Configuring the session cookie duration

Whenever an event fires, the Tracker creates a session cookie. If the cookie didn’t previously exist, the Tracker interprets this as the start of a new session.

By default the session cookie expires after 30 minutes. This means that a user leaving the site and returning in under 30 minutes does not change the session. You can override this default by setting sessionCookieTimeout to a duration (in seconds) in the argmap. For example,

{ ... sessionCookieTimeout: 3600 ... }

would set the session cookie lifespan to an hour.

Configuring the storage strategy

Three strategies are made available to store the Tracker’s state: cookies, local storage or no storage at all. You can set the strategy with the help of the stateStorageStrategy parameter in the argmap to “cookieAndLocalStorage” (the default), “cookie”, “localStorage” or “none” respectively.

When choosing local storage, the Tracker will additionally store events in local storage before sending them so that they can be recovered if the user leaves the page before they are sent.

Adding predefined contexts

The JavaScript Tracker comes with many predefined contexts which you can automatically add to every event you send. To enable them, simply add them to the contexts field of the argmap as above.

webPage context

When the JavaScript Tracker loads on a page, it generates a new page view UUID. If the webPage context is enabled, then a context containing this UUID is attached to every page view.

performanceTiming context

If this context is enabled, the JavaScript Tracker will use the create a context JSON from the window.performance.timing object, along with the Chrome firstPaintTime field (renamed to "chromeFirstPaint") if it exists. This data can be used to calculate page performance metrics.

Note that if you fire a page view event as soon as the page loads, the domCompleteloadEventStartloadEventEnd, and chromeFirstPaint metrics in the Navigation Timing API may be set to zero. This is because those properties are only known once all scripts on the page have finished executing. See the Advanced Usage page for more information on circumventing this limitation. Additionally the redirectStartredirectEnd, and secureConnectionStart are set to 0 if there is no redirect or a secure connection is not requested.

For more information on the Navigation Timing API, see the specification.

gaCookies context

If this context is enabled, the JavaScript Tracker will look for Google Analytics cookies (specifically the “__utma”, “__utmb”, “__utmc”, “__utmv”, “__utmz”, and “_ga” cookies) and combine their values into a JSON which gets sent with every event.

geolocation context

If this context is enabled, the JavaScript Tracker will attempt to create a context from the visitor’s geolocation information. If the visitor has not already given or denied the website permission to use their geolocation information, a prompt will appear. If they give permission, then all events from that moment on will include their geolocation information.

For more information on the geolocation API, see the specification.

optimizelyExperiments context

If this context is enabled the JavaScript Tracker will use the window['optimizely'].data.experiments object to create an array of context JSONs; one for each sub-object.

To see what will be captured please see the JsonSchema file com.optimizely/experiment/jsonschema/1-0-0.

optimizelyStates context

If this context is enabled the JavaScript Tracker will use the window['optimizely'].data.state object to create an array of context JSONs; one for each sub-object.

To see what will be captured please see the JsonSchema file com.optimizely/state/jsonschema/1-0-0.

optimizelyVariations context

If this context is enabled the JavaScript Tracker will use the window['optimizely'].data.variations object to create an array of context JSONs; one for each sub-object.

To see what will be captured please see the JsonSchema file com.optimizely/variation/jsonschema/1-0-0.

optimizelyVisitor context

If this context is enabled the JavaScript Tracker will use the window['optimizely'].data.visitor object to create a context JSON.

To see what will be captured please see the JsonSchema file com.optimizely/visitor/jsonschema/1-0-0.

optimizelyAudiences context

If this context is enabled the JavaScript Tracker will use the window['optimizely'].data.visitor.audiences object to create an array of context JSONs; one for each sub-object.

To see what will be captured please see the JsonSchema file com.optimizely/visitor_audience/jsonschema/1-0-0.

optimizelyDimensions context

If this context is enabled the JavaScript Tracker will use the window['optimizely'].data.visitor.dimensions object to create an array of context JSONs; one for each sub-object.

To see what will be captured please see the JsonSchema file com.optimizely/visitor_dimension/jsonschema/1-0-0.

optimizelySummary context

Unlike previously mentioned Optimizely contexts this context doesn’t attach existing in browser object, but constructs its own using only data necessary to join with exported Optimizely data.

To see what will be captured please see the JsonSchema file com.optimizely.snowplow/optimizely_summary/jsonschema/1-0-0.

We highly recommend to use this context instead of previous ones because it is has much smaller footprint and contains all necessary data.

optimizelyXSummary context

Support for OptimizelyX has been introduced in the tracker, you can have a look at the JsonSchema in com.optimizely.optimizelyx/summary/jsonschema/1-0-0 to see what is being captured.

If you’re planning on leveraging the context’s variation names, you’ll have to untick ‘Mask descriptive names in project code and third-party integrations’ in the OptimizelyX menu -> Settings -> Privacy. Otherwise, all variation names will be null.

parrable context

If this context is enabled, the JavaScript Tracker will use the window['_hawk'] object to create a Parrable context JSON.

To see what will captured, please see the JsonSchema file com.parrable/encrypted_payload/jsonschema/1-0-0.

POST support

If you set the eventMethod field of the argmap to post, the tracker will send events using POST requests rather than GET requests. In browsers such as Internet Explorer 9 which do not support cross-origin XMLHttpRequests, the tracker will fall back to using GET.

eventMethod defaults to post, other options available are get for GET requests and beacon for using the Beacon API.

The main advantage of POST requests is that they circumvent Internet Explorer’s maximum URL length of 2083 characters by storing the event data in the body of the request rather than the querystring.

The Clojure Collector and Scala Stream Collector accept events sent by POST; the Cloudfront Collector does not.

You can also batch events sent by POST by setting a numeric bufferSize field in the argmap. This is the number of events to buffer before sending them all in a single POST. If the user navigates away from the page while the buffer is only partially full, the tracker will attempt to send all stored events immediately, but this often doesn’t happen before the page unloads. Normally the tracker will store unsent events in localStorage, meaning that unsent events will be resent when the user next visits a page on the same domain. The bufferSize defaults to 1, meaning events are sent as soon as they are created.

If you have set bufferSize to greater than 1, you can flush the buffer using the flushBuffer method:

snowplow_name_here("flushBuffer");

For instance, if you wish to send several events at once, you might make the API calls to create the events and store them and then and call flushBuffer afterwards to ensure they are all sent before the user leaves the page.

Note that if localStorage is inaccessible or you are not using it to store data, the buffer size will always be 1 to prevent losing events when the user leaves the page.

Beacon API support

The Beacon interface is used to schedule asynchronous and non-blocking requests to a web server. This will allow events to be sent even after a webpage is closed. This browser interface can be used to send events by setting the eventMethod field in the argmap to beacon.

Note: the Beacon API makes POST requests.

More information and documentation about the Beacon API can be found here.

POST path

The POST path that is used to send POST requests to a collector can be change with the argmap value postPath.

postPath defaults to the standard path: /com.snowplowanalytics.snowplow/tp2

Note: changing postPath is non-standard behavior that does not conform to well-defined collector protocol.

Care must be taken to ensure that requests are redirected to the collector at the correct endpoint (normally this is /com.snowplowanalytics.snowplow/tp2).

Configuring cross-domain tracking

The JavaScript Tracker can add an additional parameter named “_sp” to the querystring of outbound links. Its value includes the domain user ID for the current page and the time at which the link was clicked. This makes these values visible in the “url” field of events sent by an instance of the JavaScript Tracker on the destination page. The enrichment process will use these values to populate the refr_domain_userid and refr_dvce_tstamp fields in Redshift for all events fired on the destination page.

You can configure which links get decorated this way using the crossDomainLinker field of the argmap. This field should be a function taking one argument (the link element) and return true if the link element should be decorated and false otherwise. For example, this function would only decorate those links whose destination is “http://acme.de” or whose HTML id is “crossDomainLink”:

{ crossDomainLinker: function (linkElement) { return (linkElement.href === "http://acme.de" || linkElement.id === "crossDomainLink"); } }

If you want to decorate every link to the domain github.com:

{ crossDomainLinker: function (linkElement) { return /^https:\/\/github\.com/.test(linkElement.href); } }

If you want to decorate every link, regardless of its destination:

{ crossDomainLinker: function (linkElement) { return true; } }

Note that the above will decorate “links” which are actually just JavaScript actions (with an href of "javascript:void(0)"). To exclude these links:

window.snowplow_name_here('crossDomainLinker', function(linkElement) { return linkElement.href.indexOf('javascript:') < 0; });

Note that when the tracker loads, it does not immediately decorate links. Instead it adds event listeners to links which decorate them as soon as a user clicks on them or navigates to them using the keyboard. This ensures that the timestamp added to the querystring is fresh.

If further links get added to the page after the tracker has loaded, you can use the tracker’s crossDomainLinker method to add listeners again. (Listeners won’t be added to links which already have them.)

snowplow_name_here('crossDomainLinker', function () { return (linkElement.href === "http://acme.de" || linkElement.id === "crossDomainLink"); });

Warning: If you enable link decoration, you should also make sure that at least one event is fired on the page. Firing an event causes the tracker to write the domain_userid to a cookie. If the cookie doesn’t exist when the user leaves the page, the tracker will generate a new ID for them when they return rather than keeping the old ID.

Configuring the maximum payload size in bytes

Because the Clojure Collector and the Scala Stream Collector both have a maximum request size, the Tracker limits POST requests to 40000 bytes. If the combined size of the events in localStorage is greater than this limit, they will be split into multiple POST requests. You can override this default using a maxPostBytes in the argmap.

The Clojure Collector can’t handle requests bigger than 64kB. The Scala Stream Collector cannot process requests bigger than 50kB because that is the maximum size of a Kinesis record.

Automatically discover and set the root domain

If the optional discoverRootDomain field of the argmap is set to true, the Tracker automatically discovers and sets the configCookieDomain value to the root domain.

NOTE: If you have been setting this manually please note that the automatic detection does not prepend a ‘.’ to the domain. For example a root domain of “.mydomain.com” would become “mydomain.com”. This is because the library we use for setting cookies doesn’t care about the difference.

This will then result in a different domain hash, so we recommend that if you have been setting this manually with a leading ‘.’ to continue to do so manually.

Configuring the cookies lifetime

Whenever tracker initialized on your domain – it will set domain-specific visitor’s cookies. By default, these cookies will be active for 2 years. You can change this duration using cookieLifetime argmap parameter or setVisitorCookieTimeout method.

snowplow_name_here("newTracker", "cf", "d3rkrsqld9gmqf.cloudfront.net", { cookieLifetime: 86400 * 31, });

or

snowplow_name_here('setVisitorCookieTimeout', 86400 * 30); // 30 days

If cookieLifetime is set to 0, the cookie will expire at the end of the session (when the browser closes). If set to -1, the first-party cookies will be disabled.

Tracking prerendered pages

Some browsers can “preload” pages while user typing URL in. These users not always end up in that page, however due page preloading tracker is initialized and loaded.

JS Tracker by default doesn’t fire events when page is preloaded, but sets callback on visibilitychange event, which fires actual event only when page starts to render.

To explicitly enable tracking for prerendered pages you can use setCountPreRendered function:

snowplow_name_here("setCountPreRendered", true);

Limiting Local Storage queue size

Because most browsers limit Local Storage to around 5mb per site, you may want to limit the number of events the tracker will queue in local storage if they fail to send. The default is a max queue size of 1000, but you may wish to reduce this if your web application also makes use local storage. To do so, you should set the optional maxLocalStorageQueueSize field of the argmap is set to your desired value (e.g. 500).

Reset Page Ping on Page View

By default the tracker will reset the Page Ping timers, which were configured when enableActivityTracking is called, as well as reset the attached Page View contexts on all future Page Pings when a new trackPageView event occurs. This is enabled by default as of 2.13.0 and is particularly useful for Single Page Applications (SPA), if you previously relied on this behaviour, you can disable this functionality by specifying resetActivityTrackingOnPageView: false in the argmap on tracker initialisation.