Segment

• A single library to include that then loads other analytics tools with easy/no code switching adding new tools

Best naming convention is OBJECT ACTION like "Event Created" and only track the flows you want not everything

Server vs Client Tracking

All libs just wrap the HTTP API

Client:

• Needed for some tools(no server-side full-story)

• Richer User Data(cookies, IP, user agent, referrer, UTMs, mobile device info)

• Automatically assigns userId and stores in local storage

Server:

• More accurate as can't be blocked by adblockers/etc

• No automatic userID, must be explicitly set

Client Usage

Every event automatically grabs page info and user agent

/* global analytics */ //eslint fix
analytics.identify([userId], [traits], [options], [callback]);

analytics.identify(id, {
  email: email,
  status: source,
});

analytics.identify(id,{
    name: user.name,
    email: user.email,
  },
  {
    integrations: {
      All: true,
      Mailchimp: false,
    },
  }
);

analytics.group(user.organization_id);

analytics.track(event, [properties], [options], [callback]);
analytics.track("Demo Recorded", {video_id});

Don't bother calling identify for anonymous client users, let Segment do its random id thing stored in local storage

npm i -D @types/segment-analytics

Server Usage

Elixir/Phoenix

Unoffical support

mix.ex

{:segment, "~> 0.2.3"}

config.exs

config :segment,
  write_key: "2iFFnRsCfi" #Segment API KEY

application.ex in the supervisor list

{Segment, Application.get_env(:segment, :write_key)}

Usage

Segment.Analytics.track(user_id, event, %{property1: "", property2: ""})
Segment.Analytics.identify(user_id, %{trait1: "", trait2: ""})

Node

const Analytics = require("analytics-node");
const analytics = new Analytics(conf.get("SEGMENT_KEY"));

analytics.track({userId: id, event: "Demo Recorded"})
analytics.identify({userId: id, traits: {email, username}})