GA4 vs Universal Analytics: What Marketers Need to Know in 2026

Universal Analytics is gone. Google shut down standard UA properties on July 1, 2024, and 360 properties followed on July 1, 2024 as well. The data collected by those properties is no longer accessible — the interfaces are offline, the APIs no longer respond, and exports you didn't pull before the deadline are lost.

If you're still comparing your current GA4 numbers to what you remember from UA, you're comparing two fundamentally different measurement systems. This guide explains the key differences, what GA4 genuinely does better, and where it still frustrates experienced analysts.

Why the Shift Happened

Universal Analytics was built in an era of desktop-first, cookie-based tracking. The model assumed: - A user = a browser cookie - A session = a period of continuous activity on a single domain - A conversion = a goal triggered by a URL, duration, or event

By 2020, this model was already strained. Users switched between devices. Safari's ITP was killing third-party and even first-party cookies. Cross-domain journeys were invisible. GDPR and CCPA made cookie-based identity legally risky.

GA4 was designed to be measurement for this environment — privacy-first, cross-platform, machine-learning-assisted, and flexible enough to work with or without cookies.

The Core Architectural Difference: Events vs. Hit Types

In Universal Analytics, data was collected as typed hit types: - Pageviews - Events (category, action, label, value) - Ecommerce hits - Social interactions - User timing hits

Each hit type had its own schema and its own place in the data model. You couldn't have a pageview and an event in the same hit. Reports were organised around this distinction.

GA4 has one data structure: events. Everything is an event. A pageview is a page_view event. A purchase is a purchase event. An ecommerce impression is a view_item event. Every event can carry up to 25 parameters.

This is a cleaner model in theory, and it makes GA4 much more flexible. But it also means you need to be deliberate about what you track — nothing gets measured "for free" the way pageviews did in UA.

Sessions and Users: What Changed

Sessions

UA sessions were time-based and ended after 30 minutes of inactivity or at midnight. GA4 sessions work similarly but with differences:

• GA4 sessions do not restart at midnight — a session that starts at 11:45 PM and runs to 12:15 AM is one session
• GA4 sessions do not restart when campaign parameters change mid-session (a UA behaviour that inflated session counts for multi-campaign users)
• GA4 sessions are tracked via a `session_id` parameter on events, not as a separate hit type

In practice, GA4 session counts are often lower than UA session counts for the same traffic — sometimes 10-20% lower. This is not a tracking gap; it's the corrected model.

Users

UA identified users primarily via the _ga cookie (client ID). GA4 adds two additional identity layers:

1. User ID — if your app has authenticated users, you can pass a hashed user ID to GA4. This ties cross-device sessions to the same person.
2. Google Signals — for users who are signed into Google and have personalisation enabled, Google can stitch cross-device sessions together using their Google Account.

GA4 also has an identity resolution hierarchy: User ID takes precedence, then Google Signals, then Device/cookie ID. The "blended" user model in GA4 is more accurate than UA for cross-device measurement, but it's also less transparent — you can't always tell how GA4 identified a user in a given session.

What GA4 Does Better

Cross-Platform and Cross-Device Tracking

GA4 was built from the start to handle web and app measurement in the same property. If your business has a website and a mobile app, GA4 lets you see the complete user journey across both surfaces in the same reports. In UA, you needed separate properties for app tracking (Firebase) and web tracking, and combining them required BigQuery.

Funnel Exploration

GA4's Exploration reports include a proper funnel analysis tool with: - Closed vs. open funnels (open funnels allow users to enter at any step) - Step-by-step segment comparisons - Elapsed time between steps - Abandonment counts at each step

UA's funnel visualisation was limited to destination goals and showed only the previous step's drop-off. GA4's funnel explorer is meaningfully better.

BigQuery Export (Free)

GA4 exports hit-level event data to BigQuery for free. UA required a 360 subscription (which cost $150,000+/year) to access BigQuery. This means you can now: - Write custom SQL queries against your raw event data - Join GA4 data with CRM or ad platform data - Build custom attribution models - Retain data beyond GA4's 14-month standard retention

The BigQuery export is one of GA4's strongest features and is underutilised by most teams.

Predictive Audiences

GA4 uses machine learning to build predictive audiences: - Purchase probability — users likely to purchase in the next 7 days - Churn probability — users likely to stop engaging in the next 7 days - Predicted revenue — expected revenue from a user in the next 28 days

These audiences sync directly to Google Ads for remarketing. They require a minimum of 1,000 conversions in the last 28 days to activate, so they're not available for all properties — but for mid-to-large ecommerce and SaaS, they're powerful.

Privacy Architecture

GA4 was built for a world of consent requirements. Consent Mode V2, server-side tagging, and the cookieless measurement modelling are all native to GA4. UA could be patched to work with some of these, but it was never designed for it.

What GA4 Does Worse (and Remains Frustrating)

Reporting Interface

This is almost universally acknowledged: GA4's reporting UI is harder to navigate than UA. Standard reports are less customisable out of the box. The Exploration reports (where the real power is) are buried a level deeper. Many common UA reports — like the Referrals report or the organic keywords breakdown — require more steps to replicate in GA4.

It gets better with familiarity, but the learning curve is real.

Sampling and Data Thresholds

GA4 applies data sampling in Exploration reports when your property exceeds certain row limits, and data thresholds (blurring of small audience segments) when Google Signals is enabled. UA also sampled, but GA4's sampling behaviour in Explorations can feel more aggressive on smaller properties.

The workaround for serious analysis is BigQuery export — but not every team has SQL skills.

No View-Level Filtering

UA properties had Views — you could have a raw data view, a filtered view (excluding internal traffic), and a test view. GA4 has one data stream per platform type and uses Data Filters at the property level instead. This is less flexible, particularly for agencies managing clients who need filtered sub-views.

Historical Data Is Gone

This is not a technical limitation of GA4 — it's a business decision Google made. UA historical data was not migrated into GA4 properties. If you didn't export your UA data before July 2024, it is gone. Year-over-year comparisons using pre-migration data are impossible unless you exported to BigQuery or a third-party warehouse first.

Making the Most of GA4 in 2026

Two years into the GA4 era, the teams getting the most value from it are: 1. Using BigQuery export for any serious data work 2. Building custom Exploration reports rather than relying on standard reports 3. Implementing User ID for authenticated users 4. Using Consent Mode V2 and accepting the modelled data 5. Auditing their event and key event configuration regularly

The teams struggling are the ones who migrated the minimum viable implementation in 2023, never audited it, and are now making decisions based on incomplete event tracking and broken conversions.

Run a GA4 health score check to see how your current implementation stacks up — it takes 60 seconds and connects directly to your property.

---

Connect your GA4 and run a health check in 60 seconds at GA4Audit.ai.