The Google Analytics 4 features pros actually use
Universal Analytics is gone and everyone's on GA4 now — using maybe 10% of it, and usually the wrong 10%. Here are the features that actually earn their keep, the ones to ignore, and the setup mistake we see on almost every account we inherit.
Universal Analytics stopped collecting data on July 1, 2023, and Google deleted the historical data for standard properties through 2024. If you're still hunting for the old reports, that's not a feature gap — that's a property that no longer exists. Everyone is on Google Analytics 4 now, whether they like it or not.
And mostly they don't like it. GA4's defaults are noisy, the data model is unfamiliar, and the reports it shows you out of the box are not the ones you want. So most teams end up using maybe 10% of the tool — and, in our experience inheriting accounts, usually the wrong 10%. They babysit the Realtime report and the Reports snapshot, both of which are close to useless for actual decisions, and never touch the parts that replaced everything good about the old setup.
Here's what actually earns its keep.
Explorations: the free pivot tables nobody opens
Explorations are the single best thing in GA4, and they're free. This is where the analysis happens — free-form tables you build yourself, funnel reports, path analysis, segment overlap, cohorts. It's the closest thing to the old custom reports, except more flexible.
If you want to know how a specific landing-page cohort moved through your funnel, where they dropped, and what they did instead, you build it here in about five minutes. The standard Reports section will never tell you that. Spend your GA4 time in Explore and almost everywhere else becomes optional.
The event model: everything is an event now
The biggest conceptual change from UA is that GA4 has no pageviews-and-sessions-first model. Everything is an event — a pageview, a scroll, a click, a form submit, a purchase. That sounds abstract until it bites you: the reason your numbers look wrong is almost always that events weren't set up to match how your business actually works.
The events that matter get promoted to key events (what UA called "goals" or "conversions"). This is where setups live or die. We routinely inherit accounts where either nothing is marked as a key event, so there's no conversion data at all, or everything is, so the conversion number is meaningless. Getting this right — a small set of events that map to real business outcomes — is 80% of a useful GA4 install.
Free BigQuery export: the one UA never had
This is the feature that genuinely justifies the migration. GA4 will pipe your raw, event-level, unsampled data into BigQuery for free. UA only offered that on the $150k/year 360 tier.
Why you'd care:
- No sampling — the in-app reports estimate once you cross certain thresholds; the raw export does not.
- You own the data. Query it, join it to CRM or revenue data, keep it forever, and never get caught flat-footed again when a platform sunsets and takes your history with it.
- It's the foundation for any serious attribution or LTV work downstream.
Turn it on today even if you have no immediate use for it. Storage is nearly free and future-you will be grateful the day someone asks a question the standard reports can't answer.
Looker Studio for dashboards people will actually read
GA4's own UI is not a reporting layer you hand to a client or a CMO. Looker Studio (free, connects natively to both GA4 and the BigQuery export) is. Build the five numbers that matter into a clean dashboard, schedule it, and stop making people log into GA4 to hunt for answers. This is the modern, no-code version of the custom-dashboard ambitions everyone had a decade ago — and it actually works.
Consent mode and modeled data: the privacy-era plumbing
Less glamorous, increasingly non-optional. With cookie consent and tracking prevention eating into measurable traffic, GA4's consent mode lets you collect privacy-respecting signals and fill the gaps with modeled conversions. You don't have to love it, but if you run paid media you need conversions reported consistently, and this is the mechanism. Set it up deliberately rather than letting it default.
What the pros ignore
Just as useful to know what to skip. The Realtime report is a fun toy and a terrible decision-making tool. The Reports snapshot and overview cards are clutter. "Users" as a headline metric is shakier than people assume once consent and cross-device modeling are in play. And the default attribution settings are rarely the ones you'd choose. None of these are where the value is.
The mistake on almost every account we inherit
If we audit your GA4 and find one problem, it's this cluster: key events not configured (or everything marked as one), internal and developer traffic never filtered out, and the whole thing left on defaults. The result is data that looks confident and is quietly wrong — which is worse than no data, because people make decisions on it.
GA4 is free, it's powerful in the places that count, and most of its reputation problem is really a setup problem. Get the event model right, turn on the BigQuery export, do your real work in Explorations, and report out of Looker Studio. That's the whole game.
If you'd rather not become a GA4 archaeologist, that's literally part of what we do — see how we handle analytics and measurement.
Can I still get my old Universal Analytics data back?
No. Universal Analytics stopped processing new data on July 1, 2023, and Google deleted the historical data for standard UA properties starting in July 2024. If you didn't export it before then, it's gone for good.
The lesson worth taking from that: turn on GA4's free BigQuery export now, so the next time a platform sunsets you still own your raw history.
Why don't my GA4 numbers match my ad platforms or my old UA reports?
Because they're counting different things. GA4 is event-based where UA was session-based, attribution windows differ, and consent mode plus modeled conversions change the totals. Ad platforms count conversions their own way too, with their own attribution.
They will never match exactly, and chasing parity is a waste of time. Pick GA4 as your source of truth for on-site behavior, understand why each system differs, and reconcile — don't expect the numbers to be identical.
Do I need to pay for GA4 360?
Almost certainly not. The 360 tier (six figures a year) is for enterprises hitting data-volume limits, sampling thresholds, or needing SLAs and higher quotas. For nearly everyone else, the free tier plus the BigQuery export covers it. Spend the money on getting the free version set up correctly first — that's where the real gap usually is.
What's the most common GA4 setup mistake you see?
Key events. We inherit accounts where either nothing is marked as a key event (so there's no conversion data at all) or everything is (so the conversion count is meaningless). Right behind it: internal and developer traffic that was never filtered out, and an account left entirely on defaults. Each one quietly corrupts the data while the reports still look confident.
Is GA4 enough on its own, or do I need a CDP or product analytics tool?
For most small and mid-market businesses, GA4 plus the BigQuery export plus Looker Studio is plenty. Add tooling when you have a specific job GA4 genuinely can't do — server-side tracking, deep product analytics, identity resolution across systems — not by default. Buying more tools before GA4 is set up correctly just gives you more places to be wrong.