DefinitionGoogle Analytics 4, Google's current analytics platform. It tracks users and events across web and apps and replaced Universal Analytics in 2023.
GA4 is the analytics tool most websites use to see where visitors come from and what they do. Everything in it is an event: a page view, a click, an add to cart, a purchase. That is a different model from the old Universal Analytics, which was built around sessions and page views, and it is why so many people who knew the old tool find the new one confusing. The mistake I see most is assuming GA4 works out of the box. It records page views and a few automatic events by default, but conversions, e-commerce data and clean channel attribution only exist if someone sets them up properly.
Why GA4 matters for SEO decisions and revenue
Google Search Console tells you how people arrive from Google. GA4 tells you what happens next, and that is where the money is. A Dutch lease platform I worked with had steady organic traffic growth for a year and no idea whether it was worth anything, because lead form submissions were not tracked as events. Once the form and the phone clicks were measured, it turned out that two blog posts produced more qualified leads than the entire pricing section, which changed where the content budget went.
For webshops, GA4 with proper e-commerce tracking shows revenue per landing page, per channel and per product category. That lets you answer the questions that actually matter: which category pages should get more SEO attention, whether the new landing page converts better than the old one, and whether the traffic from a new market is buying or just browsing. Without it you are optimising for sessions, and sessions do not pay invoices.
How to check whether your GA4 setup can be trusted
Open Admin, then Events, and look at what is being collected. If the list contains only page_view, session_start, scroll and first_visit, nothing business-specific is being measured. Then check Key events (the current name for conversions): a webshop should have purchase marked, a B2B site should have its form submission and ideally phone or email clicks. Use DebugView with the Tag Assistant browser extension to walk through a purchase or a form submission and confirm the events fire once, with the right parameters and the right value.
Compare the purchase count and revenue in GA4 against the shop’s own order admin for the same period. A gap of 5 to 15 percent is normal because of consent, ad blockers and browser restrictions; a gap of 50 percent means the tracking is broken. Check the Traffic acquisition report for a large “Unassigned” channel, which usually means UTM parameters are wrong or the referral exclusions are missing. Finally, in Admin, confirm that data retention is set to 14 months rather than the default 2, and that Google Signals and Search Console are linked. For reporting, most of my clients look at a Looker Studio dashboard fed by GA4 rather than at GA4 itself, because the standard interface is hard to read for anyone who does not use it weekly.
Common GA4 mistakes
No consent mode, or consent mode done wrong. In the EU you need consent for analytics cookies. Consent Mode v2, set up correctly, lets GA4 model the missing data. Set up wrong, you either collect nothing or you collect illegally.
Double tagging. The theme loads GA4, a plugin loads it again and Google Tag Manager loads it a third time. Every page view counts three times and bounce rate looks fantastic.
Marking the wrong events as key events. If “scroll” and “click” are key events, the conversion rate is meaningless and any Google Ads campaign optimising on it is bidding on nonsense.
Ignoring the thresholding. With Google Signals on, small reports get data withheld for privacy. If numbers do not add up in a filtered report, that is usually why.
Trusting attribution blindly. The default data-driven model shifts credit between channels in ways that are opaque. Compare it with last-click before drawing conclusions about which channel to cut.
How I use GA4 in client work
My first step on any new client is a measurement audit: what fires, what is missing, what is duplicated, and whether the numbers match the order system. I fix the setup through Google Tag Manager rather than hard-coded scripts, because it keeps the tracking independent of the theme and lets me add events without a developer. For webshops that means the full e-commerce event chain from view_item to purchase; for B2B sites it means form submissions, phone clicks, quote downloads and, where relevant, which service page the lead came from.
Then I build a Looker Studio dashboard that combines GA4 with Search Console and, for shops, the platform’s own order data. The client sees organic revenue per category, leads per landing page and the trend over time on one page, and I use the same dashboard to decide what to work on next month. GA4 on its own is a data source; the dashboard is where it becomes a decision tool.
Frequently asked questions
Is GA4 free?
Yes, for practically every business. The free version has sampling and data limits that only very large sites reach. Google Analytics 360, the paid tier, adds higher limits, SLAs and BigQuery volume, and is priced for enterprises. Exporting GA4 data to BigQuery is free in the standard version too, which is useful for anyone who wants raw event data beyond the interface.
Why does GA4 show fewer sessions than Universal Analytics did?
Sessions are defined differently. GA4 does not start a new session when the campaign source changes mid-visit, it handles midnight differently, and consent mode means some visits are modelled rather than counted. Lower session counts in GA4 are expected and not a sign of lost traffic. Compare users and purchases instead, which are more stable between the two.
Do I need Google Tag Manager to use GA4?
No, but I would not run a serious setup without it. A plain GA4 tag in the theme covers page views. Anything beyond that, such as form tracking, e-commerce events or consent handling, becomes far easier to build, test and change in Tag Manager, and it survives theme changes and platform migrations.