Learn Google Analytics

GA4 is Google's event-based analytics product that replaced Universal Analytics in 2023.

An account holds properties; a property holds data streams; a web stream gives you the G- measurement ID.

Every interaction is an event with a name and key-value parameters; GA4 has four kinds of events.

How GA4 counts users and sessions, and what makes a session or user 'engaged'.

Dimensions describe, metrics count, and each belongs to a user, session, event or item scope.

How GA4 decides that two events belong to the same person: User-ID, device ID and modelling.

Paste the Google tag snippet into <head> on every page, or load GA4 through Google Tag Manager.

Toggle automatic scroll, outbound click, site search, video, file download and form events in the web stream.

Check that events and parameters arrive as intended before you trust a single report.

Tag your campaign URLs with utm_source, utm_medium and utm_campaign so GA4 attributes traffic correctly.

Use gtag('event', name, {params}) to record interactions GA4 cannot see on its own, preferring recommended names.

Mark the events that represent success as key events so GA4 counts and attributes them everywhere.

Register your own parameters as custom dimensions and metrics, and describe users with user properties.

Handle client-side routing and multi-domain journeys so sessions and page views stay correct.

Find your way around the Reports section: snapshot, Realtime, collections, dates, comparisons and filters.

User acquisition vs traffic acquisition, and how source/medium become default channel groups.

Read the Events, Pages and screens, Landing page, Retention, User attributes and Tech reports.

Edit report cards, dimensions and metrics, build new detail reports and organise collections in the Library.

Explore queries raw event data with segments, funnels and paths; it can sample and is limited to the retention window.

Variables, Tab settings and the canvas: how the three panels of an exploration fit together.

Build a table with rows, columns and values, nest rows, pivot and filter: the workhorse of Explore.

Switch a free-form tab to donut, line, scatter, bar or geo, and turn on anomaly detection on a line chart.

Define ordered steps and see how many users complete each one, where they drop off, and how long it took.

Choose between open and closed funnels, and use breakdown, elapsed time, next action and trends to diagnose a leak.

Build subsets of users, sessions or events with scoped conditions, tell them apart from filters, and compare them with segment overlap.

Define users by an ordered series of steps, with direct or indirect succession and time limits.

Follow what users did before or after an event or page as a tree of nodes, forwards or backwards.

Inspect individual users' event timelines and delete a single user's data from there.

Group users by when they arrived or acted and measure their return, plus per-user lifetime value and engagement.

A worked example that strings reports, free form, a funnel and a segment into a repeatable weekly review.

How GA4 samples large exploration queries, how thresholding withholds rows, and what you can do about each.

Send view_item, add_to_cart, begin_checkout and purchase with an items array to light up revenue reporting.

Read the Ecommerce purchases, Purchase journey and Checkout journey reports, and item vs event revenue.

How GA4 credits key events to channels: data-driven and last-click models, lookback windows and the paths report.

Connect Google Ads for cost and conversion sharing, and Search Console for queries and organic landing pages.

Exclude your own team's and developers' hits with traffic_type rules and data filters, carefully, because filtering is permanent.

Choose 2 or 14 months of event retention, understand what it does and does not affect, and delete data on request.

A pre-launch checklist for a new GA4 property, including who gets which role.

Set consent defaults before the tag loads, update them from your banner, and know GA4's privacy controls.

Diagnose duplicate tags, self-referrals, unwanted referrals, missing campaigns, (not set) and bot traffic from their fingerprints.

Define groups of users in Admin for reporting, triggers and remarketing, and understand how they differ from segments.

Purchase probability, churn probability and predicted revenue, and the ready-made predictive audiences.

Export raw events to BigQuery daily or streaming, understand the nested schema and why it escapes sampling and retention.

Send events to GA4 from a server with an HTTP POST, to augment (not replace) browser data.