Google Analytics 4: Complete Usage Tutorial

If you opened your reports last summer and noticed a friendly little banner telling you that Universal Analytics had stopped processing hits, you were not alone. Google Analytics 4 took over the throne on July 1, 2023, and the old session-based world many of us grew up with quietly closed shop. GA4 is a different animal. Everything is an event.

Pageviews, scrolls, file downloads, video plays, even that user who rage-clicks your CTA three times in a row — they all flow into the same event stream and are scored against engagement metrics rather than bounce rate. That single design decision changes how you think about reporting, how you design your tagging plan, and how you talk to stakeholders who spent five years getting comfortable with sessions and bounce rate.

That shift trips up a lot of marketers. You log into the new property, you see fewer reports, you panic, and then you start Googling phrases like direct google analytics or direct traffic google analytics definition because the dashboard feels less explanatory than the old one. The instinct is understandable.

Universal Analytics held your hand through dozens of pre-built views — demographics, behaviour flow, content drilldown — and GA4 ships with a much shorter standard menu. What you give up in pre-built reports you get back in flexibility, but only once you learn the Explorations workspace and the event-parameter model underneath everything.

This tutorial walks you through GA4 the way a working analyst would — property setup, event tracking, exploration reports, the GA4 API, and when it actually makes sense to climb the ladder to the paid Google Analytics 360 tier. We will keep the jargon light and the screenshots in your head. By the end you should know exactly how to install a clean property, fire custom events, mark conversions, build audiences, query the Data API, and decide whether the enterprise tier is worth the six-figure cheque.

Setting up a clean GA4 property is more important than the dashboards you build later. If your data layer is messy, no amount of audience tinkering will save you. A google analytic 4 install starts with three decisions: how many data streams you need, whether you will tag through Google Tag Manager or hardcoded gtag.js, and what your default attribution window should be.

For a normal marketing site one web stream is enough; for a SaaS with both a marketing site and an in-product app you almost always want two streams under the same google analytics 4 property so user behaviour can be stitched together. Stitching is what lets you see a visitor who arrives from a Google Ad on the marketing site and later converts as a logged-in user inside the product, all on the same user_pseudo_id and ideally on the same user_id once they sign in.

An example of google analytics code snippet in its modern form is just two lines: a gtag.js loader pointing at your Measurement ID and a config call. Everything else — purchases, sign-ups, scroll depth — rides on top through events fired from GTM triggers or your application code. People sometimes try to hand-code dozens of events directly into application JavaScript and it works, but it becomes a maintenance nightmare the first time marketing wants to tweak a trigger condition.

Centralising every event through GTM means non-engineers can adjust tagging without a release cycle, and you keep an audit trail of who changed what and when. Whichever path you choose, write the choice down in your team's tagging playbook so the next analyst inherits a clear standard rather than archaeological mystery.

GA4 keeps the same three-level hierarchy as Universal Analytics: account, property, data stream. What changed is the property itself. There are no more views, and the old Reporting API endpoints have been swapped for the new google analytic api — the Data API v1 — which exposes runReport, batchRunReports, and runRealtimeReport. The Admin API handles property creation, custom dimensions, and audiences programmatically.

Anyone migrating from the old v3 Reporting API will notice that dimension and metric names changed almost everywhere. Where you once asked for ga:sessions, you now ask for sessions; ga:transactionRevenue is now purchaseRevenue; ga:pageviews is screenPageViews. It is annoying for the first afternoon and forgettable by the second.

Access roles also became cleaner. Instead of the messy old combination of property and view permissions, GA4 uses Viewer, Analyst, Editor, and Administrator. Pair those with data restrictions for cost and revenue metrics if you have stakeholders who should see traffic but not margins. Every change is logged in the Change History panel under Admin, which is genuinely useful when a number suddenly moves and three different consultants have been editing the same property.

For most teams running a google analytics 4 audit, the smart move is to script the audit against the Admin API instead of manually clicking around — it scales, and it survives the next interface refresh. Pull every property, list every conversion, dump every custom dimension, and diff the result against your tagging spec. That single nightly job catches more drift than a quarterly manual review ever will.

The event-based model is what makes GA4 powerful and frustrating at the same time. Every hit is an event with a name and up to 25 parameters. Some events are built in (page_view, scroll, click, file_download), some are recommended (sign_up, purchase, generate_lead), and the rest are custom.

Naming discipline matters: pick lowercase_snake_case and stick to it forever, because once an event name exists in a property you cannot rename it without breaking historical comparisons. Mixed casing — PurchaseComplete vs purchase_complete vs purchaseDone — is the single most common mistake we see when auditing inherited properties, and the only real fix is a long migration of stitched reports.

The event count google analytics card in the standard reports tells you how often each event fired, while engagement rate google analytics measures the share of sessions that lasted longer than ten seconds, had a conversion, or fired at least two pageviews. Engagement rate is GA4's polite replacement for the much-mourned bounce rate. Watch the engagement number per landing page; a sudden dip usually means a page slowed down, a hero image broke, or a script started crashing on mobile Safari.

Treat parameters as a finite resource. You can register up to 50 event-scoped and 25 user-scoped custom dimensions on the free tier. Reserve a few slots for future use, document every parameter you send, and avoid the urge to attach a brand-new parameter to every campaign — most of them are better expressed as values inside an existing parameter.

Reports in GA4 are deliberately sparse. The library on the left contains Life Cycle (Acquisition, Engagement, Monetization, Retention) and User (Demographics, Tech). Anything fancier — funnels, paths, cohorts, segment overlap — lives inside Explorations. That move was controversial but it actually pushes marketers toward better questions: instead of clicking around hoping a chart will explain itself, you have to build the chart that answers your question. Save the explorations that prove their worth, share them as templates, and prune the rest aggressively.

Realtime is still there, with the last 30 minutes plotted live, useful for verifying a campaign just went out. DebugView is the unsung hero — connect it to your testing device and you can see every event fire with all its parameters before you ship the tagging. Any expert google analytics consultant will spend more time in DebugView than in any other screen during a build. It is also where you catch the rare but spectacular bug of an event firing twice because both a hardcoded gtag call and a GTM tag are listening to the same DOM trigger.

The Library section lets you customise the left-hand navigation per role. Hide reports your team does not need, surface the ones they live in, and treat the report library as the front page of your internal analytics product. Most teams underuse this feature and it is the cheapest way to make GA4 feel friendly to non-analysts.

Audiences in GA4 are powerful because they update in near-real time and they can be exported straight to Google Ads for remarketing without a separate Ads-side rule. You can build them out of any combination of events, parameters, user properties, and predictive metrics like purchase probability. A useful default set covers engaged_users (engaged sessions ≥ 2), cart_abandoners (added_to_cart but no purchase in 7 days), and high_value_visitors (top 10% of lifetime engagement time). Build them once, give them clear names, and document the logic inside the description field so future analysts do not have to reverse-engineer the rules.

Predictive audiences require enough conversion volume to train Google's model — usually around 1,000 positive and 1,000 negative samples in the past 28 days. Smaller sites will see predictive metrics greyed out and that is fine; you still get rule-based audiences which cover ninety percent of marketing needs. Pair audiences with funnel exploration to see exactly where each segment falls off your journey, then decide whether the right intervention is a new email campaign, a UX fix, or a retargeting push.

Be careful with overlapping audiences. Stack too many small segments and you will end up bidding against yourself across Google Ads campaigns. Use the segment overlap exploration to visualise which audiences share users, and prune the duplicates before media budget gets fragmented across half-overlapping lists. Quarterly audience hygiene is one of the simplest wins in any GA4 account.

The hardest call most growing teams face is when to graduate from free GA4 to paid GA360 or whether to look across the fence at Adobe. The answer almost always comes down to three things: sampling, freshness, and contractual support. Free GA4 standard reports are unsampled, but Explorations and the Data API start sampling above the 10-million-event monthly ceiling. If your business model lives or dies on freshly accurate revenue numbers, that ceiling will hurt. Smaller brands can sail past it for years and never feel the pinch.

An enterprise analytics migration google analytics 360 project usually pays for itself when paid media spend tops about $250k a year, because unsampled data plus a 360-day attribution window changes how you bid. The google analytics 360 cost is negotiated, but the public-facing entry point sits around $150k per year and scales by event volume. You also unlock sub-properties — virtual splits of the same data into different reporting buckets, ideal for agencies serving multiple brands inside one account.

Adobe Analytics is the other major option and it remains the strongest choice for organisations already invested in the Adobe Experience Cloud — Target, AEM, Campaign. Adobe's calculated metrics and workspace pivots are unmatched, but the learning curve is steep and the licence cost can dwarf GA360. Most pure-play digital businesses end up on GA4 or GA360 simply because the Google Ads integration is frictionless. Choose based on where your media spend lives, not the marketing brochure.

One small thing that catches first-time admins out: the google analytics 4 logo is no longer the orange and yellow swirl you remember from Universal Analytics. The new mark — a blue gradient bar chart — appears in Google Marketing Platform, in the GA4 property header, and in any embed cards.

If you are writing internal documentation or training decks, use the official brand asset rather than re-saving a screenshot, because the older logo can confuse stakeholders into thinking your screenshots are out of date. It is a tiny detail and it disproportionately matters when you are presenting to executives who judge polish on first impression.

Branding aside, treat the property as a living data product. Build a small style guide that pins down event names, conversion definitions, and which custom dimensions are reserved. Stakeholders will trust the numbers more when they understand the rules behind them. The guide does not need to be a novel — a single Google Doc with a table of events, parameters, and owners covers ninety percent of what new team members need to ramp up.

Make every change reviewable. Anyone with Editor access can break tagging in seconds, but the Change History panel gives you a paper trail. Pair that with quarterly access reviews and you have a lightweight governance model that scales from a two-person startup to a multinational without changing tools.

You do not need a six-figure tool to run analytics well. Most teams will live happily on the free GA4 tier for years, especially if they invest in tagging hygiene, write down their event taxonomy, and treat their BigQuery export as the canonical source of truth rather than the in-product reports. The in-product reports are for stakeholders; BigQuery is for the analytics team. Once that division is clear, every other tool decision becomes simpler — Looker Studio for executive dashboards, Hex or Mode for analyst notebooks, reverse-ETL into your CDP for activation.

Promote yourself up the maturity curve gradually. Spend a sprint on a clean property setup, another on event tracking, another on audiences and funnel exploration, and only then start scripting the google analytic api or evaluating GA360. By the time the question of paying for the enterprise tier actually matters, you will have the volume numbers, the attribution evidence, and the political backing to make the case. Until then, do the boring work — name your events well, audit your tags monthly, and build the explorations your colleagues actually open.

And whatever you do, write things down. Future you will be grateful for every tagging note, every audience definition, every conversion description that explains why a metric is calculated the way it is. The teams that get the most value out of GA4 are not the ones with the fanciest dashboards. They are the ones whose analysts can answer a question in five minutes because the property is documented, the events are clean, and the warehouse query they need is already written and saved. Build that boring foundation first, and the impressive dashboards become almost easy to deliver.