What is Google Analytics 4 (GA4)?

Google Analytics 4 (GA4) is the current version of Google’s free web and app analytics platform, which replaced Universal Analytics in July 2023. It records what visitors do on a website as a stream of events (page views, scrolls, clicks, form submissions, purchases), ties those events to users and sessions, and reports where visitors came from, what they did and which of them completed the actions the business cares about. It is the system of record for traffic and conversions on most small and mid-sized business websites.

GA4 is also where many businesses’ reporting quietly went wrong. The move from Universal Analytics changed the data model, the interface, the default retention period and the meaning of familiar numbers, and a large share of sites were migrated with a default tag and nothing else. The result is a property that counts page views and little that a business decision could rest on. Setting it up properly is a half-day of work; leaving it on defaults costs every month afterward.

How GA4 models data

Universal Analytics was built around sessions and page views with separate hit types for events, transactions and so on. GA4 makes everything an event. A page view is an event named page_view; a purchase is an event named purchase with parameters for value and items; a form submission is whatever event you define. Users and sessions are built from the event stream rather than being the primary unit.

• Events are the actions. GA4 collects some automatically (page_view, session_start, first_visit), offers “enhanced measurement” for common ones (scroll, outbound click, file download, video engagement, site search), and lets you define the rest.
• Parameters are the details attached to an event: page location, form name, button text, product value. Custom parameters have to be registered as custom dimensions or metrics before they appear in reports, which is the step most setups miss. Our guide to tracking custom parameters and A/B tests in GA4 walks through it.
• Key events (called conversions until 2024) are the events you mark as the ones that matter. Marking a form submission as a key event is what makes conversion rate reporting possible.
• Users are identified by a device cookie by default, optionally by a user ID you provide or by Google signals. The same person on a phone and a laptop is two users unless something links them.
• Sessions are groups of events from one user within a window of inactivity (30 minutes by default). Unlike Universal Analytics, a session does not restart at midnight or when the traffic source changes.

What GA4 reports

• Acquisition: where users and sessions came from, by channel (organic search, paid search, direct, referral, organic social, email), source and medium, and campaign when UTM parameters are present.
• Engagement: which pages and screens were viewed, which events fired, engagement rate and average engagement time, which replaced bounce rate as the headline measure of whether visitors stayed.
• Monetization: e-commerce revenue, items and purchase funnels, when the site sends e-commerce events.
• Retention and demographics: returning users over time, and age, gender, interest and location data where Google signals is enabled and thresholds are met.
• Explorations: free-form tables, funnels, path analysis and segment overlap, which is where the questions standard reports cannot answer get answered.
• Advertising: attribution across channels for key events, with data-driven attribution as the default model.

The settings most sites leave wrong

A GA4 property with the default configuration is missing several things that cannot be recovered later, because analytics data is not retroactive.

• Data retention is two months by default. Explorations can only look back that far. Change it to 14 months in the property settings on day one. Standard reports keep aggregate data longer, but the detailed data is gone.
• No key events defined. Without them, the property knows nothing about leads or sales, and every acquisition report is just traffic.
• Internal traffic not excluded. Your own visits, your agency’s visits and your developer’s test submissions all count until a filter is defined and activated.
• Search Console and Google Ads not linked. Linking Search Console adds query reports; linking Ads imports cost and lets campaigns optimize toward key events.
• Enhanced measurement counting the wrong things. The form interaction events it fires automatically are unreliable on many WordPress form plugins, and the outbound click event counts links to your own booking subdomain as exits. Define real events instead.
• Cross-domain tracking not configured. A booking tool or checkout on another domain starts a new session with a new source, so the conversion is credited to “referral” from your own tool. Add every domain in the tag’s configuration.
• Consent mode absent or misconfigured. With a consent banner that blocks the tag until accepted, a large share of visitors are never measured; consent mode sends modeled, cookieless pings instead so the totals are not simply missing.

GA4 on WordPress

The tag can be installed three ways, and the choice determines how much can be measured later. Pasting the gtag snippet into the theme measures page views and enhanced events and nothing else without code changes. A plugin (Site Kit, MonsterInsights and others) does the same with a settings screen and some extra weight on every page. Google Tag Manager installs one container and lets events be defined, tested and changed without touching the site, which is how we set up every client site: form submissions, phone clicks, booking-button clicks and scroll depth are defined once in the container, pushed to GA4 and to the ad platforms from the same trigger, and changed without a deploy. The form plugin matters too; one that fires a clean JavaScript event on successful submission is far easier to track than one that only redirects to a thank-you page or, worse, reloads the page.

Attribution on WordPress has one more piece: the form itself should capture the source, medium and campaign of the visitor who submitted it, so the lead in your inbox or CRM says where it came from. GA4 can tell you campaigns produced 40 form submissions; it cannot tell your sales team which 40. Our guide to capturing UTM parameters in WordPress forms covers the mechanism.

Reading GA4 without misreading it

• Numbers are modeled and thresholded. Reports with demographic or signals data hide rows below a privacy threshold, and some metrics are estimated from samples once a property is large. Small differences between reports are normal.
• Direct traffic is unexplained traffic. Visits with no referrer: typed URLs, bookmarks, links from apps and documents, and anything whose attribution was lost. A rise in direct traffic is more often a tracking gap than a surge in brand awareness.
• Engagement rate is not the inverse of the old bounce rate. An engaged session lasted 10 seconds, had two page views or fired a key event. Many single-page visits that would have been bounces now count as engaged.
• Attribution is a choice. Data-driven attribution shares credit across the touchpoints before a key event; the acquisition reports default to last click for sessions. The same conversion can be credited differently in two reports on purpose.
• GA4 and Search Console disagree by design. Search Console counts clicks on Google; GA4 counts sessions that loaded the tag. GA4 will always show fewer organic sessions than Search Console shows clicks.

Privacy and alternatives

GA4 was built for a world with consent requirements, cookie restrictions and users who block trackers, and it compensates with modeling, consent mode and server-side tagging. For many small businesses that is enough, and the price (free, with the data available to Google) is acceptable. Businesses with stricter requirements, or who want simpler reports without Google’s involvement, use privacy-focused alternatives such as Plausible, Fathom or Matomo, which count visits and a few events with no cookies and no consent banner, at a monthly cost and with less depth. Running one of those alongside GA4 for a quarter is a reasonable way to see how much GA4 is missing after consent.

What to look at each month

A working GA4 property answers four questions in ten minutes: how many key events happened, which channels and campaigns produced them, which landing pages produced them, and how all of that compares to the previous period. Everything else in the interface is there for when one of those four numbers moved and you need to know why. If the property cannot answer the four questions, the fix is configuration, not more reports.

Configuring GA4, Tag Manager and form tracking so that every lead is counted and attributed is part of our marketing and analytics services, and it is the first thing we check before any conversion rate optimization work. If your analytics shows traffic but cannot show leads, book a discovery call and we will set it right.