How Google Analytics 4 Works: Events, Reports, and Limits

You open Google Analytics 4 looking for a report you have used for years, and it is gone. Bounce rate vanished from the main panel. Goals became "key events". The menu looks like it was reorganized by someone who never sat next to a marketing manager.

The frustration is common, and it has a precise cause: GA4 is not an upgrade of Universal Analytics. It is a different tool built on a different collection logic. Anyone who tries to use it the old way gets lost, and concludes, unfairly, that "the new Analytics is bad".

How Google Analytics 4 works, in short: GA4 records every user interaction, page view, click, scroll, purchase, as an event with parameters. Those events feed standard reports and custom explorations, let you flag conversions (key events), and can be exported to BigQuery at no cost.

With that settled, three pieces remain: what actually changed, which reports deserve your time, and where the tool's limits are. Let's take them in that order.

Why is Google Analytics 4 so different from Universal Analytics?

Universal Analytics was designed in the late 2000s, when "a visit to the website" was the unit that mattered. Everything revolved around sessions and pageviews. GA4 was built for a different landscape, multiple devices, apps, cookie restrictions, and swapped the core unit: everything is now an event, and the session became just a grouping of events.

This is not a niche tool. According to W3Techs, Google Analytics runs on 47.6% of all websites and holds an 83.2% share of the traffic-analysis tool market (August 2026 data). It is the web's de facto measurement standard, which turns this change of logic into everyone's problem.

What changed	Universal Analytics	Google Analytics 4
Unit of measurement	Session and pageview	Event with parameters
Quality metric	Bounce rate	Engagement rate
Conversions	Session-based goals	Key events (per event)
Raw data retention	Configurable, could never expire	14 months max on the free tier
BigQuery export	Paid version only	Included in the free tier

How does the GA4 event model work?

In GA4, everything is an event. A page_view is an event. A button click, a scroll to 90% of the page, a purchase: events. Each one carries parameters, details such as value, currency, page name, or click source.

Events come from four sources: automatically collected ones (page_view, session_start), enhanced measurement events (scrolls, outbound clicks, site searches, enabled with one click), recommended events (purchase, generate_lead, with names standardized by Google), and custom events, which you create for whatever is left.

A concrete example: an online store logs the purchase event with the parameters value: 89.90, currency: USD, and the list of items bought. That design is what lets GA4 calculate revenue per channel and average order value without goals tied to thank-you page URLs, as in Universal Analytics.

The single most important configuration step is flagging the events that represent business outcomes, purchase, lead, booking, as key events. Without it, acquisition reports show traffic but never tell you which channel drives results. And then the tool becomes an expensive visit counter.

Which GA4 reports are worth your time?

GA4 ships with dozens of reports, and you will never open most of them. For day-to-day marketing, four cover almost everything:

• Traffic acquisition: which channel brings sessions and key events. Add the campaign secondary dimension to separate campaigns from brand traffic.
• Pages and screens: where people land, what they read, and where engagement dies.
• Key events: conversion counts by source, the metric that should open every marketing meeting.
• Monetization: for e-commerce, revenue by item, list, and coupon, straight from event collection.

When the ready-made reports fall short, explorations take over: funnels, path analyses, and cohorts you assemble yourself. A funnel exploration can show, for instance, that 68% of users who start checkout abandon at the shipping step, the kind of finding no standard report hands you.

One combined-use case: a fashion e-commerce brand running Meta Ads and Google Ads opens traffic acquisition and sees Meta leading in sessions. Comparing key events per session flips the picture, paid search converts nearly three times better. Without that second metric, the next budget shift would go to the wrong channel.

If you are wondering what to track beyond GA4, cross it with your paid traffic metrics and the KPIs for each funnel stage, Analytics measures the middle of the journey, not the whole business.

What are GA4's limitations?

This is the part nobody covers during onboarding. GA4 has real limits, and knowing them prevents wrong decisions built on incomplete data.

Short data retention. On the free tier, the event data behind explorations lasts at most 14 months, and many properties still sit on the initial 2-month setting without knowing it. Standard aggregated reports never expire; but any custom analysis comparing this year with last year may simply have no data. Check that setting today.

Data thresholding. With Google Signals on, GA4 hides report rows with few users to protect privacy. On lower-traffic sites, metrics "disappear" with no clear warning, and the parts stop adding up to the total.

Sampling in explorations. Above 10 million events per query, explorations run on a sample rather than the full dataset. Most SMBs never hit this; for large operations, custom numbers become approximations.

Opaque attribution. The data-driven model splits credit across channels with machine learning but never explains the weights. With healthy conversion volume it works reasonably well; with few key events per month, treat attribution numbers as estimates, not truth.

Processing latency. Data can take 24 to 48 hours to consolidate. Making a Monday-morning budget call on weekend numbers is asking for trouble.

How to use Google Analytics 4 without getting lost: a 6-step routine

1. Set data retention to 14 months (Admin, Data settings, Data retention).
2. Define 3 to 5 key events that represent real outcomes, purchase, qualified lead, booking, and ignore the rest.
3. Review enhanced measurement: confirm the automatic events make sense for your site.
4. Standardize UTMs across every campaign; without them, acquisition fills up with "unassigned".
5. Build one funnel exploration for your main flow and review it monthly.
6. If long history matters, turn on the BigQuery export now, it only stores data from activation onward.

With that in place, GA4 answers mid-funnel questions well. What it cannot do is close the loop on the business: to know whether the operation is healthy, you still need to cross behavior with cost and revenue, calculating ROAS and being data driven in practice require data that lives outside Analytics.

Frequently asked questions about Google Analytics 4

Is Google Analytics 4 free?

Yes, the standard version is free and covers the vast majority of companies. There is a paid tier (Analytics 360) for very large volumes, with higher retention and processing limits. If you have to ask what 360 costs, you probably do not need it.

How do I install Google Analytics 4 on my site?

Create the property in Analytics, get the measurement ID (G-XXXX), and install the tag: directly in the code with gtag.js, through Google Tag Manager (the most flexible route), or via a plugin such as Site Kit on WordPress. Verify collection in the Realtime report before trusting the data.

What is the difference between GA4 and Universal Analytics?

UA measured sessions and pageviews; GA4 measures events with parameters. That changes metrics (engagement instead of bounce), conversions (key events instead of goals), and reports. UA stopped collecting data in July 2023, direct comparisons between the two will never match.

What are key events in GA4?

They are the events you flag as important business outcomes, formerly called conversions. They feed acquisition reports and serve as bidding signals in Google Ads. Choosing the right key events decides whether GA4 measures results or just motion.

Does GA4 replace a BI tool?

No. It collects and organizes behavior on your site and app, but it does not join that data with your CRM, finance system, or cost spreadsheets. Full business analysis needs a layer on top, BigQuery, a BI tool, or an AI analytics platform.

Measuring got harder, and reading became the bottleneck

GA4 is a competent collector and a hostile reader. The tool records nearly everything that happens on your site; pulling answers out of it is what costs time and training. Our position: learn the minimal routine above, resist the urge to configure everything, and accept that deep analysis may not belong inside GA4 at all.

That is where Sherlok comes in: connect GA4 alongside Meta Ads, Google Ads, and your CRM, then ask questions in plain language, "which channel brought the highest-ticket customers this quarter?", and get the analysis ready, without ever opening an exploration.