Most explanations of what is GA4 stop at “it replaced Universal Analytics.” That’s true and useless. The real answer is that GA4 threw out the measurement model you spent a decade learning and replaced it with a different one that answers different questions — and if you keep asking it Universal Analytics questions, it hands back numbers that look wrong because you’re reading them wrong. Understanding what GA4 actually is means understanding the shift from sessions to events, why that shift makes some of your old reports impossible, and where it quietly fails you as an SEO. Get that, and the interface stops feeling hostile.
The one-sentence definition, then the real one
The one-sentence version: GA4 is Google’s event-based web and app analytics platform, and it’s now the only version Google offers. The real version is more useful. Universal Analytics thought in sessions — a visit was the atomic unit, and everything hung off it. GA4 thinks in events — every page view, scroll, click, video play, and file download is a discrete event with its own parameters, and a “session” is just an event GA4 assembles after the fact. That single design decision explains nearly every quirk you’ll hit. Once you internalize that GA4 counts things that happen rather than visits that contain things, the confusing parts start to make sense.
Why the metrics changed (and which comparisons are now meaningless)
The renamed metrics aren’t cosmetic. “Bounce rate” is gone in its old form because a bounce assumed a session-based worldview; GA4 gives you engagement rate instead — the share of sessions that lasted 10+ seconds, fired a conversion, or had 2+ page views. “Goals” became key events. “Users” is now split across active users, total users, and new users, and the default reports show active users, which is why your headline number can look lower than Universal Analytics ever did.
The trap is year-over-year comparison. If you screenshot a 2021 Universal Analytics report next to a 2026 GA4 report and declare traffic “down 30%,” you’re often just measuring two different definitions of a user against each other. They were never the same metric. The honest move is to treat your GA4 install date as a new baseline and stop comparing across the model boundary — trend GA4 against GA4, not against a ghost.
What GA4 collects for free the moment it’s installed
Install the base tag and GA4 starts recording enhanced measurement events automatically, no extra code:
- page_view — every page load, including history changes on single-page apps
- scroll — fires at 90% page depth
- click — on any outbound link leaving your domain
- view_search_results — when a URL contains a search query parameter
- file_download — clicks on PDFs, docs, and other file types
- video_start / video_progress / video_complete — for embedded YouTube players
This matters because a lot of “I need a developer to track X” turns out to be already tracked. Before you write custom events, open Admin → Data Streams → Enhanced Measurement and see what’s on. The gap between what people think GA4 captures and what it captures out of the box is enormous.
Setup that actually gives you usable data
Installing GA4 is a ten-minute job; installing it well is what separates usable data from a graveyard of default reports. Three settings do most of the work, and Google buries or defaults them badly:
- Data retention: defaults to 2 months. Change it to 14 months (the max) in Admin → Data Retention or your exploration reports go blind past 60 days.
- Link Google Search Console: without this, GA4 has zero organic keyword or query context — more on that failure below.
- Mark your key events: a form submit or purchase is just another event until you flag it as a key event. Unflagged, it never shows up as a conversion.
Use Google Tag Manager rather than hardcoding the tag if you can — it makes adding custom events later a config change instead of a deploy. And define custom events for the actions that actually matter to your business (a quote request, a plan selection), because GA4’s automatic events don’t know what your money moment is.
Explorations: where GA4 actually gets powerful
The standard reports are deliberately shallow — Google wants you in the Explore section, which most people never open. This is the real answer to “what is GA4 good for.” Explorations let you build free-form tables, funnels, path analysis, and cohort reports by dragging dimensions and metrics onto a canvas. A funnel exploration shows you exactly where users drop between landing page, add-to-cart, and checkout. A path exploration shows the actual next-page sequence after someone lands on a blog post — invaluable for internal-linking decisions. If you only ever use the left-nav reports, you’re using maybe a fifth of GA4. Explorations are the fifth that answers real questions.
Audiences, segments, and predictive metrics
GA4 can build audiences — reusable groups like “visited pricing but didn’t convert in 7 days” — that both segment your reports and, if you’ve linked Google Ads, feed remarketing. On properties with enough conversion volume, GA4 also exposes predictive metrics: purchase probability, churn probability, and predicted revenue, generated by Google’s machine-learning models. These are genuinely useful for prioritizing which visitors to chase, but honest caveat: they need a meaningful volume of positive and negative examples over the prior 28 days to activate, so most small sites will see them greyed out. Don’t build a strategy around a metric your traffic can’t unlock yet.
BigQuery export: the feature that removes GA4’s ceilings
Here’s the upgrade nobody mentions in a beginner guide: GA4 offers a free daily BigQuery export of your raw event data. This is the single biggest reason GA4 is more capable than Universal Analytics ever was for serious analysts. In the standard interface you hit data thresholding (Google hides rows to protect user privacy) and cardinality limits (high-variety dimensions get bucketed into “(other)”). Export to BigQuery and those disappear — you query the raw, unsampled, un-thresholded event rows with SQL. You can join GA4 data to your CRM, your ad spend, anything. It’s a config toggle in Admin → BigQuery Links, and the export itself is free within BigQuery’s generous sandbox tier. Turn it on today even if you won’t query it for months, because it only captures data from the day you enable it forward — there’s no backfill.
A worked micro-example: reading a real drop
Say a blog post’s traffic falls in your GA4 landing-page report. The amateur reads “sessions down 40%” and panics. Walk it properly. First, is the drop in active users or a metric-definition artifact? Confirm the date range doesn’t straddle a data-retention wall. Next, open a free-form exploration, set the dimension to Session default channel group, and filter to that landing page. Now you can see the drop is entirely in “Organic Search” while “Direct” held steady — so it’s a ranking problem, not a site-wide tracking break. Cross-reference the linked Search Console report: impressions flat, clicks down means your ranking position slipped, not your search demand. That four-step read — user metric, retention, channel split, Search Console cross-check — turns a scary chart into a specific diagnosis. GA4 tells you that organic fell; it can’t tell you which query fell, which is the exact gap we’ll close next.
What GA4 will never tell you about SEO
This is the honest limitation an SEO has to hold onto. GA4 does not know which keyword produced a visit. Google stripped organic keyword data from analytics years ago (“not provided”), and GA4 didn’t bring it back. It can tell you organic search sent 500 users to a page; it cannot tell you 300 came from one query that just slipped from position 3 to 8. For that you need Search Console query data, and for competitive keyword intelligence you need a dedicated index.
This is exactly where a tool like SEO Rocket earns its place beside GA4 rather than replacing it. GA4 is your ground-truth for on-site behavior and conversions; SEO Rocket handles the layer GA4 is blind to — AI keyword research on real Ahrefs data, competitor content-gap analysis, and rank tracking that shows which query moved and why. The workflow that scaled a portfolio past 1,000,000 ranking pages always paired the two: GA4 to confirm what visitors did after they arrived, a real keyword index to understand how they found you in the first place.
The quirks that generate support tickets
A few behaviors look like bugs and aren’t. Processing delay: standard reports can lag 24–48 hours, so “today’s” numbers are always incomplete — use the Realtime report for the last 30 minutes and stop refreshing the main reports. Data thresholding: when you apply demographic dimensions to a low-traffic segment, GA4 hides the rows entirely to prevent re-identification, so totals stop reconciling. Attribution recalculation: GA4 reassigns conversion credit retroactively as more of a user’s journey completes, so a number you screenshotted Monday can legitimately differ Wednesday. None of these are errors; they’re the privacy-and-event model doing its job. Knowing that saves you a week of chasing phantom problems.
Frequently asked questions
Is GA4 free?
Yes. GA4 is free for the vast majority of sites, including the BigQuery export within BigQuery’s sandbox tier. The paid tier (GA4 360) exists for enterprises needing higher data limits, faster processing, and SLAs, but you’ll know if you need it — most businesses never will.
Do I still need Google Search Console if I have GA4?
Absolutely, and they’re not interchangeable. GA4 measures what happens on your site; Search Console measures how you appear in Google’s results — impressions, queries, positions, click-through rate. GA4 can’t see any of that. Link them so GA4’s Search Console reports light up, and keep both open.
Why don’t my GA4 numbers match my old Universal Analytics numbers?
Because they’re measuring different things. Session definitions, the user metric shown by default, and how bounces are counted all changed with the event model. Don’t reconcile them — set your GA4 install as a fresh baseline and trend forward from there.
How long does GA4 take to show data?
Realtime appears within minutes; standard reports typically finalize in 24–48 hours. If a brand-new property shows nothing after a day, check that the tag actually fires (use the Realtime report or Tag Assistant) before assuming the data is lost.
Where to start this week
So, what is GA4 in the end? It’s a privacy-first, event-based measurement system that’s more powerful than what it replaced and less intuitive — a fair trade once you stop fighting it. Do three things this week: switch data retention to 14 months, link Search Console, and toggle on the BigQuery export so future-you has raw data to query. Then open one exploration and rebuild a report you actually care about from scratch. Pair GA4’s behavioral truth with a real keyword index — SEO Rocket for the query and competitor layer GA4 can’t see — and you’ve got the full picture: how people find you, and what they do once they arrive.
