What Is GA4? Google Analytics 4 Explained and Why It Is Worth Using

GA4, short for Google Analytics 4, is Google's free measurement tool for websites and apps. I have set up analytics for client sites since 2012, and one lesson keeps repeating itself. When you configure GA4 properly, you see what your website actually does for the business. When you configure it badly, it gives you pretty charts that mean very little.

This article does not compare analytics tools. Instead, I focus on how GA4 itself thinks: the event model, the core reports, key events, BigQuery and consent mode. My goal is simple. By the end, you should know what you are looking at in the interface and which numbers deserve your trust.

What is GA4 and what does it do?

GA4 is the version of Google Analytics that records every user interaction on a website or app as an "event". Page views, scrolls, form submissions and purchases all follow the same event structure. As a result, you can measure traffic sources, content performance and conversion paths with one consistent data model.

In practice, GA4 answers three questions. Where did the visitor come from? What did they do on the site? Did they take a step that matters to your business? However, the quality of those answers depends entirely on your setup. For example, if you never define the form submission as its own event, GA4 can only tell you how many people viewed the form page.

Also, GA4 is not a standalone tool. It sits at the centre of an ecosystem, and you can connect it to Google Ads, Search Console and BigQuery. That is why your measurement plan should serve advertising and SEO decisions too, not only the analytics team.

Why did Universal Analytics shut down, and what changed with GA4?

Universal Analytics stopped processing new data in standard properties on 1 July 2023. Google's reasoning was clear. The old model did not fit a world with fewer cookies and people who switch between several devices. On top of that, its session and pageview structure could not merge with app data.

GA4 turned that structure around. The old version had separate "hit types" for pageviews, events and ecommerce transactions. Now everything lives under one roof as events and parameters. In other words, GA4 records a page view and a purchase with exactly the same logic.

The difference that surprises my clients most is bounce rate. In the old version, any single page visit counted as a bounce. In GA4, bounce rate is the share of sessions without engagement. So comparing old reports with new ones one to one usually misleads you.

How does the event based data model work?

Every record in GA4 is an event, and every event carries parameters. For example, the "page_view" event sends the page URL and title as parameters. The "purchase" event carries the order value, the currency and the list of items. Reports simply aggregate these events and their parameters.

In practice, the strength of this model is flexibility. You can define your own events and attach any parameter you need. For instance, if you add a "service" parameter to a quote request event, you can report on which service attracts the most enquiries.

But flexibility has a price. To see a parameter in the standard reports, you must register it as a custom dimension. If you skip that step, GA4 still collects the data, but the reports never show it. This is the most common frustration I hear in the field: "the data goes in, but I cannot find it".

• Event: an action the user takes, such as a page view or a form submission.
• Parameter: a detail of that event, such as the page URL or the order value.
• User property: a lasting attribute of the user, such as membership type.
• Custom dimension: the definition that makes a parameter visible in reports.

Which events does GA4 collect automatically?

GA4 collects some events without any work on your side. For example, "first_visit" and "session_start" arrive automatically. On top of that, there is a setting called enhanced measurement. When you switch it on, GA4 records several common interactions without extra code.

1. Page views (page_view).
2. Scrolls, which fire when a user reaches roughly 90% of the page depth (scroll).
3. Outbound link clicks (click).
4. Site search results (view_search_results).
5. Progress on embedded YouTube videos.
6. File downloads (file_download).
7. Form interactions (form_start and form_submit).

However, I have one warning here. Automatic form tracking does not work reliably with every form setup. I often see double counts or missing counts, especially with forms that submit without a page reload or with third party form tools. So I always track business critical forms with a dedicated event, often through Google Tag Manager.

You can also switch each enhanced measurement option on or off individually. For example, if your site has no embedded videos, turning video tracking off keeps the reports clean. For site search, check the query parameter name, because GA4 misses searches when your site uses something other than the usual names. In short, enhanced measurement gives you a good start, but it does not replace a measurement plan.

How does GA4 calculate sessions and engagement rate?

A GA4 session starts with the "session_start" event and, by default, ends after 30 minutes of inactivity. Google's help page on sessions also defines an engaged session. It is a session that lasts longer than 10 seconds, includes a key event, or has at least two page or screen views.

Engagement rate is the share of engaged sessions among all sessions. Bounce rate is simply the opposite. So if engagement rate is 60%, bounce rate is 40%. This definition is much fairer than the old single page logic. After all, someone who reads a blog post for three minutes and then leaves no longer counts as a failed visit.

On the other hand, this metric still needs interpretation. If the goal of a page is a phone call, a visitor might tap the number within five seconds and leave. In that case, mark the call click as a key event, and GA4 will treat the session as engaged. I cover practical ways to lower bounce rate in my article on reducing bounce rate on a business website.

Where are the core reports, and what does each one tell you?

The GA4 interface looks messy at first, but the logic is simple. The "Reports" section in the left menu holds ready made summaries. Meanwhile, the "Explore" section holds the custom analyses you build yourself. For daily monitoring, most businesses need only a handful of reports.

My advice: during the first month, focus only on traffic acquisition and the events report. That way you watch both your channels and the health of your tracking at the same time. I explain the broader logic of reading reports in how to read a digital marketing report.

When do you need explorations?

Standard reports answer "what happened". Explorations let you dig into "why it happened". They include templates such as free form, funnel exploration, path exploration, segment overlap and cohort analysis. For example, a funnel exploration shows how many people you lose at each step between the product page and checkout.

Path exploration also works backwards. Start from a quote form submission and trace which pages people read before they sent it. When I plan content, this analysis often gives me clearer hints than a keyword tool does.

However, explorations have two limits. First, they depend on your data retention setting, so event level data older than that window disappears from explorations. Second, sampling can kick in on high volume properties. For those reasons, I verify the data from another angle before I base a critical decision on a single exploration. You can find out how to adapt funnel thinking to your site in my guide on building a conversion funnel.

Why does GA4 now call conversions "key events"?

In early 2024, Google renamed "conversions" in GA4 to "key events". The word "conversion" now belongs to Google Ads, where it describes the measurement that drives ad optimisation. So in GA4 you mark events that matter to your business as key events. Then you import the ones you want for bidding into Google Ads as conversions.

This split feels confusing at first, but it makes sense. Not every important event suits ad optimisation. For example, "pricing page view" can be a valuable key event for analysis. But if you make it a primary conversion in Google Ads, the bidding system will chase page visits instead of sales.

Therefore, I suggest you keep two separate lists: a broad list of key events in GA4, and a narrow list of Google Ads conversions that sit close to real revenue.

Which events should you mark as key events?

Your choice of key events depends on your business model. The rule is simple: every event you mark should sit close to a real business outcome. I cover the full goal setting process, including macro and micro conversions, in my guide on how to set website conversion goals. Here I only summarise the practical GA4 choices.

• Ecommerce: purchase as the primary event; begin_checkout and add_to_cart for analysis.
• Service business: quote form submissions, phone clicks and WhatsApp clicks.
• SaaS: trial sign ups, demo requests and upgrades to a paid plan.
• Content site: newsletter sign ups, file downloads and deep reading sessions.

Also, assign a value to each key event whenever you can. Here is an example calculation. If 20% of quote requests turn into sales and the average sale is worth 5,000 US dollars, you can give each form a value of 1,000 US dollars. As a result, you start comparing channels in money rather than in raw counts.

What do you gain by linking GA4 to Google Ads?

When you link your GA4 property to a Google Ads account, the benefit flows both ways. On one side, ad clicks, costs and campaign names appear in GA4 reports. On the other side, you can import GA4 key events into Google Ads as conversions and use GA4 audiences for remarketing.

But there is a common mistake here. Some advertisers track the same action with both the Google Ads tag and an imported GA4 conversion, then set both as primary. Google does not automatically deduplicate these two separate sources. The result is inflated conversion counts and a bidding strategy that learns the wrong lesson. That is why I pick exactly one primary source for each action.

If you want your campaigns and budget managed together with measurement, take a look at my Google Ads management service. For a quick return check, you can use the ROAS calculator.

Why do channel grouping and UTM tags matter so much?

GA4 sorts every session into default channel groups based on source, medium and campaign. For example, typical groups include Organic Search, Paid Search, Organic Social, Email and Referral. This grouping relies on the UTM parameters in your links.

Here is the problem, then. An Instagram link without UTM tags often shows up as Referral, or even as Direct. So the contribution of social media quietly disappears from your reports. The same thing happens with email newsletters and WhatsApp shares.

The fix is disciplined tagging. Agree on a small, lowercase vocabulary for source and medium. For example, always write "instagram", and never switch between "Instagram" and "ig". Instead of typing links by hand, generate them with the UTM builder, which removes most typing errors.

When do you need the BigQuery export?

The BigQuery link sends raw GA4 event data to a data warehouse in Google Cloud. You need it when the interface starts to hold you back. For example, if you want event level data older than your retention window, or you want to join GA4 data with CRM records, BigQuery becomes almost essential.

According to Google's BigQuery Export help page, standard properties have a daily export limit of 1 million events, while streaming export has no such limit. If you keep exceeding the limit, Google may pause your daily export. The export feature itself costs nothing on a standard property. However, storage and query costs in BigQuery still show up on your Google Cloud bill.

My own practice is simple. I switch the link on from day one on every property with serious ecommerce volume or ad spend. BigQuery does not backfill history; it only collects data from the day you connect it. So even if you do not need it today, that raw data may prove very valuable a year from now.

How does consent mode affect GA4 data?

Consent mode controls how Google tags behave for users who decline cookies. Google's consent mode documentation describes two implementations. In basic mode, tags do not load at all until the user grants consent. In advanced mode, tags send cookieless pings without identifiers while consent is missing.

Consent mode version 2 added two parameters, "ad_user_data" and "ad_personalization". These parameters now play a decisive role for ad features that involve users in the European Economic Area.

I recommend you design your cookie banner together with your legal adviser, because that part is a legal decision. My technical note is this: when consent rates are low, GA4 user and session counts sit below your real traffic. So do not panic when a report "drops" right after a banner change. Check the consent rate first.

How do data retention and data thresholds change your reports?

On standard properties, you can set GA4 to keep event level data for either 2 months or 14 months. Still, many setups run on the 2 month default. This setting does not affect standard reports. However, it directly limits how far back you can analyse in explorations. That is why one of my first steps on any new property is switching retention to 14 months.

Data thresholds are a separate matter. When Google signals are active, or when user groups are small, GA4 may hide some rows to prevent identification. If you see a warning icon in the corner of a report, a threshold applies.

As a practical fix, you can change the reporting identity to suit your needs. Also, a wider date range often brings hidden rows back. Small sites run into this a lot, so check for a threshold warning before you conclude that data has gone missing.

What are custom dimensions and audiences for?

A custom dimension turns a parameter you send into something you can use in reports. For example, add an "author" or "topic" parameter to your blog posts. Then you can see which topics drive the most key events. This lets you build your content calendar on data instead of guesswork.

Audiences, in addition, are groups of users who meet certain conditions. For instance, you might define "viewed the pricing page but did not submit a form". You can then use that audience for remarketing in Google Ads. I explain where this fits in the funnel in my article on strengthening your sales funnel with remarketing.

Keep one thing in mind, though. Most audiences do not fill up retroactively; they start collecting members from the moment you create them. So it makes sense to define the audiences you will need early. Also, small audiences can hit minimum size limits on the ads side. In short, preparing audiences weeks before a campaign saves you time.

How should you set up GA4 on an ecommerce site?

For ecommerce, the value of GA4 depends on sending the recommended ecommerce events in full. Google suggests fixed names for these events, such as view_item, add_to_cart, begin_checkout, add_payment_info and purchase. If you use those names, the monetization reports fill up on their own. If you invent different names, those reports stay empty.

I check three things. First, the purchase event must always carry an order number (transaction_id), because GA4 uses it to drop duplicate purchase events. Second, tax and shipping values should follow one consistent rule. Third, every event should state its currency explicitly.

On the other hand, if checkout moves to another domain, you need to review your domain settings so the session does not split on return. Otherwise, GA4 credits sales to the payment page instead of ads or organic search. If you want ecommerce tracking handled together with your sales strategy, see my ecommerce consulting service.

What happens if you set up GA4 without a measurement plan?

A measurement plan is a simple document where you write down in advance which question each event answers. If you skip it, you usually end up at one of two extremes. Either you define no custom events and GA4 just counts traffic. Or you turn every click into an event and the reports become meaningless noise.

In practice, the plan I use fits on one page. The left column lists the business question, the middle lists the event name and parameters, and the right column says whether it counts as a key event. For example, "which service attracts the most quotes?" maps to a "generate_lead" event with a "service" parameter. That way everyone on the team uses the same names.

So review this document once a year. Websites change, and forms, buttons and goals change with them. If you let the plan go stale, half of your reports may track buttons that no longer exist a year later.

The GA4 setup mistakes I see most often

When I take over accounts, I keep finding the same mistakes. Most of them look small from a technical angle, but they distort reports badly. The list below shows what I check first when I audit a GA4 property.

1. Double loading: the tag loads twice, once in the theme code and once in Tag Manager. The result: double page views.
2. Your own payment or booking domain appears as a referral. GA4 then credits sales to the payment provider.
3. Nobody filters internal traffic from the office, the agency or developers.
4. Data retention stays at 2 months.
5. Test orders mix with live data.
6. A key event points to the wrong event, for example a form page view instead of the form submission.

To catch most of these, the Realtime report and DebugView are enough. Send a test form from your own device and confirm that the event arrives exactly once. It sounds basic, yet many teams never run this test. Repeat it after every major site change, new form or payment switch. The check takes me about half an hour, while months of broken data cost far more.

What are the advantages of using GA4?

If I sum up the business case, GA4 offers several concrete advantages. First, it is free, and a standard property covers the needs of most small and mid sized businesses. Second, it integrates natively with Google Ads, so you can share audiences and conversions in a few clicks.

Third, the event model lets you combine web and app data in one property. Fourth, the BigQuery export is available on standard properties too; in the old version only the paid tier offered it. Fifth, metrics such as engagement rate describe visitor behaviour more fairly.

You can also connect GA4 to Search Console. This puts organic queries and landing pages side by side with analytics data. I explain Search Console itself in my Search Console guide. In short, the biggest advantage of GA4 is that, set up well, it brings your marketing decisions into one shared measurement language.

What are the limits and drawbacks of GA4?

To be honest, GA4 is not perfect. Teams used to the old version still find the interface hard. Even simple questions sometimes require a custom report or an exploration. Moreover, because of consent refusals and ad blockers, GA4 never matches your server logs or CRM exactly.

That is why I present GA4 to clients as a consistent trend indicator, not as the complete truth. For the exact number of orders, you look at your shop system or your accounting. GA4 tells you which channels and content brought those orders.

You should also factor in thresholds, sampling and retention. When these limits start to hurt, the answer is usually BigQuery plus a reporting layer. So you can often close the gaps with extra tools, but that comes with a cost in skills and time.

Which KPIs should you use to read GA4 data?

GA4 offers hundreds of metrics, but a few are enough for decisions. I usually look at sessions by channel, engagement rate, key event count, key event rate and revenue where available. Together, these five metrics show the traffic, the quality and the outcome of each channel.

For example, if a channel brings many sessions but a low engagement rate, the targeting or the landing page does not fit. On the other hand, a channel with few sessions and a high key event rate is a strong candidate for more budget.

I explain which metric belongs to which goal in my article on digital marketing KPIs. What matters most is tracking the same metrics with the same definitions every month, so you can actually see change.

What should you do in the first 30 days of a new GA4 property?

I follow the same order on every property I set up or take over. This sequence gives you a reliable baseline by the end of the first month.

• Day one: load the tag from a single source, switch retention to 14 months and define internal traffic.
• First week: test critical events, mark key events and assign values.
• Week two: link Google Ads and Search Console, and switch on BigQuery.
• By week three, share the UTM vocabulary with your team and fix untagged campaigns.
• Week four: pull the first monthly report and compare it with CRM or order data.

That last step matters a lot. If the gap between GA4 and real sales stays within a reasonable range, your tracking is healthy. If the gap is large, one of the mistakes above is usually at work. I suggest you repeat this comparison every quarter. Log the gap in a sheet each month, and you will spot tracking problems much earlier.

Conclusion: getting real value out of GA4

GA4 is a powerful tool, but it does not create value on its own. The value comes from the events you define, the key events you choose and disciplined tagging. Once you understand the event model, the reports start to make sense.

My advice is to start with a small but correct measurement plan. Three or four key events, a clean UTM system and 14 months of retention give most businesses a solid base. Then you go deeper with BigQuery and custom reports as needs grow.

If you want measurement handled together with your SEO strategy, visit my SEO consulting page, or send me your questions through the contact page.