Rails Local Analytics

Simple, performant, local analytics for Rails. Solves 95% of your needs until your ready to start taking analytics more seriously using another tool.

Out of the box the following request details are tracked:

• day
• total (count per day)
• url_hostname (site)
• url_path (page)
• referrer_hostname
• referrer_path
• platform (ios, android, linux, osx, windows, etc)
• browser_engine (blink, gecko, webkit, or nil)

It is fully customizable to store more details if desired.

Screenshots

Installation

# Gemfile
gem "rails_local_analytics"

Add the following migration to your app:

bundle exec rails g migration CreateAnalyticsTables

# db/migrations/..._create_analytics_tables.rb

class CreateAnalyticsTables < ActiveRecord::Migration[6.0]
  def up
    create_table :tracked_requests_by_day_page do |t|
      t.date :day, null: false
      t.bigint :total, null: false, default: 1
      t.string :url_hostname, null: false
      t.string :url_path, null: false
      t.string :referrer_hostname
      t.string :referrer_path
    end
    add_index :tracked_requests_by_day_page, :day

    create_table :tracked_requests_by_day_site do |t|
      t.date :day, null: false
      t.bigint :total, null: false, default: 1
      t.string :url_hostname, null: false
      t.string :platform
      t.string :browser_engine
    end
    add_index :tracked_requests_by_day_site, :day
  end

  def down
    drop_table :tracked_requests_by_day_page
    drop_table :tracked_requests_by_day_site
  end
end

Add the route for the analytics dashboard at the desired endpoint:

# config/routes.rb
mount RailsLocalAnalytics::Engine, at: "/admin/analytics"

Its generally recomended to use a background job (especially since we now have solid_queue). If you would like to disable background jobs you can use the following config:

# config/initializers/rails_local_analytics.rb
RailsLocalAnalytics.config.background_jobs = false # defaults to true

The next step is to collect traffic.

Recording requests

There are two types of analytics that we mainly target:

• Site level analytics
  • Stored in the table tracked_requests_by_day_site
• Page level analytics
  • Stored in the table tracked_requests_by_day_page

Your controllers have to manually call RailsLocalAnalytics.record_request. For example:

class ApplicationController < ActionController::Base
  after_action :record_page_view

  private

  def record_page_view
    return if !request.format.html? && !request.format.json?

    ### We accept manual overrides of any of the database fields
    ### For example if you wanted to track bots:
    site_based_attrs = {}
    if some_custom_bot_detection_method
      site_based_attrs[:platform] = "bot"
    end

    RailsLocalAnalytics.record_request(
      request: request,
      custom_attributes: { # optional
        TrackedRequestsByDaySite.name => site_based_attrs,
        #TrackedRequestsByDayPage.name => {},
      },
    )
  end
end

If you need to add more data to your events you can simply add more columns to the analytics tables and then populate these columns using the :custom_attributes argument.

Some examples of additional things you may want to track:

• Bot detection
  • Bot detection is difficult. As such we dont try to include it by default. Recommended gem for detection is crawler_detect
  • One option is to consider not tracking bots at all in your analytics, just a thought
  • You may not need to store this in a new column, one example pattern could be to store this data in the existing platform database field
• Country detection
  • Country detection is difficult. As such we dont try to include it by default.
• Users or organizations
  • You may want to track your users or another model which is a core tenant to your particular application

Performance Optimization Techniques

There are a few techniques that you can use to tailor the database for your particular needs. Heres a few examples:

• If you drop any database columns from the analytics tables this will not cause any issues. It will continue to function as normal.
• url_hostname column
  • If you wont ever have multi-site needs then you can consider removing this column
  • If storage space is an issue you may consider switching to an enum column as the number of permutations is probably something that can be anticipated.
• referrer_host and referrer_path columns
  • Consider just storing "local" or nil instead if the request originated from your website
• platform and browser_engine columns
  • Consider dropping either of these if you do not need this information

Usage where a request object is not available

If you are not in a controller or do not have access to the request object then you may pass in a hash representation. For example:

RailsLocalAnalytics.record_request(
  request: {
    host: "http://example.com",
    path: "/some/path",
    referrer: "http://example.com/some/other/path",
    user_agent: "some-user-agent",
    http_accept_language: "some-http-accept-language",
  },
  # ...
)

Deleting old data

By default all data is retained indefinately. If you would like to have automated deletion of the data, you might use the following example technique:

class ApplicationController
  after_action :record_page_view

  private

  def record_page_view
    # perform other logic and call RailsLocalAnalytics.record_request

    TrackedRequestsByDayPage.where("date < ?", 3.months.ago).delete_all
    TrackedRequestsByDaySite.where("date < ?", 3.months.ago).delete_all
  end
end

Page Performance Tracking

We dont do any page performance tracking (request/response time, etc), this gem only specializes in analytics.

If you are looking for a simple performance tracking solution, I highly recommend the gem inner_performance

Development

Run server using: bin/dev or cd test/dummy/; rails s

Testing

bundle exec rspec

We can locally test different versions of Rails using ENV['RAILS_VERSION']

export RAILS_VERSION=7.0
bundle install
bundle exec rspec

Credits

Created & Maintained by Weston Ganger - @westonganger

Imitated some parts of active_analytics. Thanks to them for the aggregate database schema idea.