Universal Track Manager, also known as UTM, is a gem to track your visitors and their pageloads. You can use it to track Urchin Tracking Module (UTM) parameters, and the fact that these two things abbreviate to the same letters is play-on-words.
You can use Universal Track Manager to track simple information like browser, IP address, http referrer, and your inbound UTM (Urchin Tracking Module) parameters. Because UTM parameters are used most commonly with advertising, we also refer to tracking your UTM parameters as "ad campaigns" or just "campaigns".
In particular, this Gem can be used to solve the common first-land problem where UTM parameters are present only in the first page of a user's visit, but not available naturally a few steps later when the event you want to track happens (see 'UTM Hooks')
Visits are a lot like Rails sessions; in fact, this Gem piggybacks off of Rails sessions for tracking the visit. (You will need to have a session store set up and in place.)
However, visits are not identical to sessions. More than one visit can have been created from the same session.
A session will have only one visit at a time. If a new visit event happens within an existing session, like the user returns in the same browser the following day, the old visit gets evicted from the session and a link between the newly created visit and old visit is maintained in the visits table.
It is important to understand there are many different data points can could possibly be collected in today's web traffic. You should use this gem at your own discretion, and you can choose either more or less data capturing, as well more or less data integration with your users.
The reason I underscore this point is that the safest data is anonymized data.
The general Rails best practice for the last decade has been to keep only session data with the user, but anonymize identifying data (like IP, behavior, browser), which could be used to identify individual users.
Please see the section 'Granularity of Data Capture' to understand the many different levels of (non-)privacy you may choose as the website operators.
In any country or region where a privacy law like the GDPR or California Consumer Privacy Act is in effect, getting informed consent to track this information is just one part of what you must do to comply with the law.
Most privacy laws regulate the usage, storage, transmission, and removal of this data once you are retaining it in your database as well.
You will note that most old privacy policies talk about much of this data being stored in "log files." This gem takes the data retention farther and stores the data into the database. (So you should modify your privacy policy accordingly.)
By using this gem you are responsible for complying with the appropriate laws and regulations subject to you.
Please familiarize yourself with the concepts above before installing.
-
add
gem 'universal-track-manager'
to yourGemfile
-
To track only the default utms (utm_campaign, utm_medium, utm_source, utm_term, utm_content), use the generator like so:
rails generate universal_track_manager:install
If you also want to track Facebook and Google clicks IDs, append the parameter --add
with a list of field names separated by COMMA.
rails generate universal_track_manager:install --add=fbclid,gcllid
To REPLACE the default list of tracked parameters, use --only
rails generate universal_track_manager:install --only=abc,def
The default utm_campaign would not be tracked in the above case
Of course, you can also use --only
to specify a subset of the UTM fields like so:
rails generate universal_track_manager:install --only=utm_source,utm_campaign,utm_medium
This will create a schema migration for the new tables that UTM will use (look for db/migrate/00000000000000_create_universal_tracking_manager_tables.rb. see 'Name Conflicts' below if any of these tables already exist in your app.)
- In your ApplicationController, add:
include UniversalTrackManagerConcern
- Notice tha the installer has created this file for you in
config/initializers/universal_track_manager.rb
UniversalTrackManager.configure do |config|
config.track_ips = true
config.track_utms = true
config.track_user_agent = true
config.campaign_columns = 'utm_source,utm_medium,utm_campaign,utm_content,utm_term'
# config.track_referrer = false
end
- Extensible Tracking
As of Version 0.7.0, you can now extend the UTM parameters to include any paramater with data for your website's inbound traffic. fbclid
or glic
are example of inbound parameters from Facebook and Google, respectively. This is useful if you are running advertising that brings people to your Rails site and the ad platforms are sending you traffic with specific, custom tracking parameters you want to keep track of.
To customize, modify the comma-separated config.campaign_columns
in the initializer above.
For optimization and speed, a unique SHA will be automatically generated from all of the combined columns. This is indexed at your database to make the lookup very fast.
For this version I have added a counter onto th visits table.
IF YOU ARE UPGRADING from a version prior to 0.7.5, you must
class AddCountToVisits < ActiveRecord::Migration[6.1]
def change
add_column :visits, :count, :integer, default: 1
end
end
All new visits start a 1, the next subsequent visit will be +1 from its parent visit.
- minor release removes byebug
- removes "-staged" files from the install generator after they are created
- changes 'param_list' to separate options 'add' (augment the default list) or 'only' (replace it)
- IMPORTANT: If you are upgrading, you must create migrations to make the following modifications from a pre-0.7.1 install
- add field
sha1
to campaigns (string) - add and index also called
sha1
(name: "index_campaigns_on_sha1") also on campaigns - this index field should be the index of ALL of the fields you want to track Here's a quick example:
class AddUTM071ChangesToUtmTables < ActiveRecord::Migration[6.1]
def change
add_column :campaigns, :sha1, :string
add_index(:campaigns, [:utm_campaign, :utm_term, :utm_content, :utm_source, :utm_medium], name: :index_campaigns_on_sha1)
end
end
- note that the limits on many of the other strings have changed to 255
(sorry I didn't keep release notes before this)
Configuration options can be set in
config/initializers/universal_track_manager.rb
which will automatically be created for you when you run the install generator. Set any option to false to disable tracking.
Note that in this section we are talking about the Urchin Tracking Module (also, non-coincidtally, abbreviated as UTMs). These appear on your inbound links as GET parameters and look like:
utm_source
utm_campaign
utm_medium
utm_term
utm_content
For example, a typical inbound link with fully coded UTM parameters might look like so:
When a visitor first lands on your site, the UTMs from that first land are assocaited to the visit (we will call these the first-land UTMs). From here, your visitor might click to a second page, or a third page, before signing up or logging in. By this time, the UTMs are no longer stored in the URL parameters because the user has clicked several pages into your website.
As the web operator, you want to know which of your advirtising campaigns are leading to which sign-ups or logins. One common approach is to store the UTMs in the session until the user signs up or logs in.
Using this gem, we do not store the UTMs in the session, but instead the session has a link to the visits
table that can used later (upon the event where you want to capture the UTMs).
If you want to grab your UTMs them from Universal Track Manager object, support is already biult-in easily from any controller:
currrent_track.campaign
You can also fetch and store the currrent_track.campaign_id
in your foreign table, which will allow you to aggregate your tracked events to your the inbound traffic sources. Remember, this Gem will not do this for you as you must associate this to your application's own unique needs. Please note that nil is nil in the database and empty string is empty string. if campaign parameters are attached to inbound links as empty string, they will create distinct records from those where no parameter was passed
UTM will create tables named browsers
, campaigns
, visits
. If you already have tables named like this, you will want to 1) edit the generated files after step 2, and 2) add an override for the UniversalTrackManager objects, like so:
/initializers/universal_track_manager.rb
// specify that this gem should use the table name `web_visits` instead of `visits`
UniversalTrackManager::Visit.class_eval do
self.table_name = :web_visits
end
// and/or specify that this gem should use the table name `web_browsers` instead of `browsers`
UniversalTrackManager::Browser.class_eval do
self.table_name = :web_browser
end
// and/or specify that this gem should use the table name `ad_campaigns` instead of `campaigns`
UniversalTrackManager::Campaign.class_eval do
self.table_name = :ad_campaigns
end
(If you already have models named Visit or Browser, don't worry: the scope of the UniversalTrackManager:: objects will prevent collisions with model-only objects in your app.)
-
UTM will set a session variable named
visit_id
in your Rails session. If you already use a session variable with the same name, please override by: -
TODO: implement this
This Gem has been designed to allow the developer to flexibly choose the granularity of personal identification options, retention and use of the data you collect (see below).
In particular, this Gem seeks to popularize and inform the Ruby community, as well as inform the broader landscape, on the granularity choices presented to a modern website in today's day and age.
While the Rails session is typically where you might store information that could individually identify the user (user or account id), the visits table is, by default, detached from the session information so that when the session information is deleted, your visit does not contain a way to easily re-identify the individual, thus making your visitors somewhat anonymous. Since a reverse-engineering can be done from other places where you might store identifably information, this system cannot gaurantee to be fully anonymized unless you take extra steps to flush, aggregate, and purge your visits table in an anonymizing way (not natively implemented). The native implementation does a reasonable job at detaching the visitors from the identifying information, thus providing partial anonymization.
Essentially, with all this granularity and security options, you have 6 levels of privacy you can choose from:
-
Most private: Don't use this gem.
-
Somewhat private: Track user visits, but use this Gem's NO_SESSION flag [NOT YET IMPLEMENTED]
-
Less private: Track user visits and link them to the Rails session, but keep the existing security concepts around the Rails session (regularly purge old records from the table when they are older than 2 weeks). When you do this, if you Rails app is compromised by a hacker (both database access and your SECRET_BASE_KEY), your hacker will have access to both the identifably information (in the session) and non-identifiable table (in the visits table) for only the people who logged in within the last 2 weeks. Furthermore, if you want to associate some of your vital information (like your UTMs), with your users once they log-in or sign-up, you must implement this yourself and you may implement it by using the hooks provided by the Gem.
4A. Barely private (bad idea from security perspective; terrible idea from privacy perspective): Track user visits and link them to the Rails session, but don't expire or purge your sessions. (not recommended)
4B. Barely private (good idea from security perspective; best idea from privacy perspective): Track user visits and link them to the Rails session, expire the sessions as expalined in #3 above, and use the long-cookie approach for an extra layer of identificiation.
-
Hardly private (good idea from security perspective; better idea from privacy perspective): Just like 4B, use the
visits
from this gem, the Rails session with expiry, and possibly a long-cookie. As well, grab some or all of the information out of thevisits
table for storage elsewhere on an as-need basis. For example, you might track only the user agent (browser) and ad campaign (UTMs) but not the IP address. Since we assume that the browser doesn't change within a single visit (by definition, it can't), you don't need this gem to look at the browser (just dorequest.user_agent
). But you can use this gem to grab the UTMs from a previsouly stored visit, explained below. -
No privacy (good idea from a security perpsective; bad idea from a privacy perspective): Just like 4B, use the
visits
from this gem, the Rails session with expiry, and possibly a long-cookie. As well, create a foreign key from some other table directly to the visits table, and associate any time a user logs-in or signs-up to their visit details.
A visit also stores the first and last time the visitor came during that visit. However, since visits may share sessions, a visit is made unique by any of:
• new session
Even within the same Rails session, a visit can be defined unique by:
• new logical day* • new or different browser • new or different IP address
??? • new or different viewport size (unless the viewport size is a reverse )
(help is appreciated!)
- fix the install generator
- track the visit IP address
- track the user agent
- track the UTMs and stash them in an assocition
-
carry forward the UTMs when evicting the visits - add switch to turn off IP tracking
- track the http referrer ( optional extension )
- invalidate the old visit if next day
- optionally exclude specific IPs from tracking (like internal IP addresses)
- track viewport screensize
- an optional long-cookie feature to drop a long-lived cookie into the visitor's browser that work separately from the Rails session
- track gclid
- track fbclid
- An optional extension to build in-house geolocation by IP address, or rely on an external service for geolocation by IP address and associate the user's looked-up location with their visit information
- Anonymized geolocation, to let you look-up IPs in order to geolocate users, but not store the actual IP addresses themselves.
- A switch to track the user before or after the controller action has rendered. Since the tracking adds a small overhead to each request, tracking after the controller has rendered makes your page respond faster for the user. But if you track before you render, you can use optionally use the tracked information to personalize, customize, or target your website to respond uniquely to the visitor.
- A way to override the session variable named :visit_id
This is gem is tested with Appraisal, Rspec-rails, and rails-controller-test (for testing integration hooks with the Rails controllers)
Before running specs, setup using
(tested with Ruby 2.6.5)
First, please create a symlink for the migrations to run properly from db/migrate (even though they are really in spec/dummy/db/migrate)
ln -s ../spec/dummy/db/migrate/ db/migrate
Then delete rm db/test.sqlite3
if it already exists in the repo.
Next, setup appraisal & bundle install with:
gem install appraisal
appraisal install
appraisal rails-6-1 rake dummy:db:migrate RAILS_ENV=test
The gem has the same footprint accross all versions of Rails, and even though Appraisal will test against different versions of Rails, it will use the same database to do so.
to run specs in all versions of Rails (see Appraisals
file)
appraisal rake spec
To run the specs in only Rails 6.1, you would run
appraisal rails-6-1 rake spec
To rebuild the Appraisal Gemfiles (in gemfiles/
) use
appraisal install
- Make sure your use case or implementation is documented.
- Make sure it is tested.
- Make sure the specs are passing.