Key takeaways
- Google Analytics started as Urchin — a paid web analytics product acquired by Google in 2005 and made free, which changed the industry overnight
- Universal Analytics (2012) introduced the measurement protocol and user-centric tracking — a significant architectural improvement over the original session-based model
- GA4 is not an upgrade to Universal Analytics — it is a completely different product built on a different data model, event schema, and collection methodology
- The July 2023 Universal Analytics sunset forced the largest forced platform migration in digital analytics history
- Understanding this history explains why GA4 works the way it does — and why migrating from UA to GA4 required rebuilding implementations rather than updating them
In this article
Most people using GA4 today have no idea where it came from. They know it replaced Universal Analytics, they know it works differently, and they know the migration was painful. But the reasons GA4 works the way it does — the event-based model, the session definition, the user identification approach — all trace back to decisions made over 20 years of Google Analytics history.
Understanding that history makes you a better analytics practitioner. It explains why certain things work the way they do, why certain limitations exist, and where the platform is likely heading.
Urchin — the origin (1995–2005)
Google Analytics did not start at Google. It started as Urchin — a web analytics product built by a company called Urchin Software Corporation, founded in San Diego in 1995.
Urchin was a server-side log analysis tool. In the mid-1990s, web analytics meant processing server access logs — the raw records that web servers kept of every HTTP request. Urchin parsed these logs and generated reports on visits, page views, referrers, and browser types. It was sophisticated for its time, and it became one of the leading commercial web analytics products of the late 1990s and early 2000s.
The shift from server-side log analysis to client-side JavaScript tracking happened gradually through the early 2000s. JavaScript page tags — small scripts that fired on page load and sent data to a collection server — gave analysts data that server logs could not: what happened after the page loaded, which links users clicked, how long they stayed. Urchin adopted JavaScript tracking and became the product that Google eventually acquired.
Google acquires Urchin (2005)
In April 2005, Google acquired Urchin Software Corporation. The acquisition price was never disclosed publicly. What happened next changed the web analytics industry permanently.
Google made the product free.
Before 2005, enterprise-grade web analytics software cost thousands of dollars per month. Urchin itself was priced at several hundred dollars per month for hosted access. When Google relaunched it as Google Analytics in November 2005 — free for any website with under 5 million pageviews per month — it immediately made professional analytics accessible to every website owner on the internet.
The decision to make Google Analytics free was strategic, not charitable. Web analytics data helped Google understand how people used the web — which informed its search and advertising products. The data Google collected through Analytics helped improve ad targeting, which was Google's primary business. Free analytics for publishers meant more data for Google's ad machine.
The launch was so popular that Google Analytics was invitation-only for the first several months. The servers could not handle the demand from millions of websites wanting free analytics overnight.
Google Analytics Classic (2005–2012)
The original Google Analytics — sometimes called GA Classic or GA2 — was essentially Urchin rebranded and hosted by Google. The tracking methodology used a JavaScript snippet (ga.js) that set a first-party cookie and sent pageview data to Google's collection servers.
The data model was session-based. A session was a series of page views from a single browser within a 30-minute window. Visitors were identified by a cookie stored in the browser. The fundamental unit of analysis was the session — how many sessions, from which sources, viewing which pages.
GA Classic introduced concepts that analytics practitioners still use today: bounce rate, goal completions, traffic sources, and the channel groupings that attributed sessions to organic, direct, referral, and paid sources. It also introduced the concept of Goals — configuring specific pages or events as conversion points.
The limitations of GA Classic were significant by modern standards. It was entirely session-based — cross-session user journeys were difficult to analyse. It had no concept of user identity across devices. Mobile app tracking was not supported. The data model assumed a simple web world that was becoming rapidly more complex.
Universal Analytics (2012–2023)
In 2012, Google launched Universal Analytics — a significant architectural upgrade to the platform. The key improvements over GA Classic:
Universal Analytics introduced the User ID — the ability to assign an authenticated identifier to a user and stitch their behaviour across devices and sessions. For the first time, analytics practitioners could see that the same person browsed on mobile, converted on desktop, and came back a week later on a work laptop.
It also introduced the Measurement Protocol — an HTTP API that allowed data to be sent to Google Analytics from any device or server, not just browsers running JavaScript. Server-side events, offline conversions, and IoT device data could all be sent to the same analytics property.
Custom Dimensions and Metrics gave analysts the ability to attach business-specific data to sessions and users — membership tier, subscription type, content category — in a way that GA Classic could not support.
Universal Analytics ran successfully from 2012 through to its sunset in 2023 — an 11-year run. It became the analytics platform that an entire generation of digital marketers and analysts built their careers on.
The App + Web experiment (2019)
In 2019, Google launched a beta product called App + Web — an attempt to create a unified analytics property that combined web and mobile app data. This was the direct precursor to GA4.
App + Web used Firebase Analytics as its foundation — the event-based mobile analytics platform that Google had acquired with Firebase in 2014. Rather than the hit-type model of Universal Analytics, App + Web used a pure event model where every interaction — including page views — was an event with parameters.
The reception was mixed. The interface was unfamiliar, reports were missing compared to Universal Analytics, and the data model required a fundamentally different way of thinking about tracking. Many analytics teams dismissed it as incomplete and continued with Universal Analytics.
GA4 — a complete rebuild (2020–present)
In October 2020, Google officially renamed App + Web as Google Analytics 4 and announced it as the future of Google Analytics. The message was clear: Universal Analytics was legacy, GA4 was the path forward.
GA4 is built on three foundational principles that differ from Universal Analytics:
Everything is an event
In Universal Analytics, there were hit types — pageview hits, event hits, transaction hits, social hits. Each had a fixed structure. In GA4, everything is an event with a name and parameters. A page view is an event called page_view. A purchase is an event called purchase. This unification makes the data model more flexible but also requires a completely different tracking implementation.
Privacy-first by design
GA4 was designed in an environment of increasing privacy regulation — GDPR, CCPA, the death of third-party cookies, and ITP on Safari. It uses modelled data to fill gaps where consent is not given, relies less on cookies for user identification, and was built with the assumption that perfect data collection is no longer possible.
Cross-platform from the ground up
Universal Analytics had separate products for web and apps that could be linked but never truly unified. GA4 unifies web and app data in a single property from the start — using the same event schema, the same user identification methods, and the same reporting interface regardless of whether the data came from a website or a mobile app.
The Universal Analytics sunset (2023)
In March 2022, Google announced that Universal Analytics would stop processing new data on July 1, 2023. For GA360 (the enterprise paid version), the cutoff was July 1, 2024.
This was not a deprecation — it was a hard shutdown. Properties that had not migrated to GA4 simply stopped collecting data. Historic UA data remained accessible for a period but was eventually deleted.
The migration was the largest forced platform transition in digital analytics history. Hundreds of millions of websites had to rebuild their tracking implementations — not update them, rebuild them. The data models were incompatible. Universal Analytics eCommerce schema did not map to GA4 eCommerce. UA custom dimensions did not map to GA4 custom dimensions. UA goals did not automatically become GA4 conversions.
Teams that had invested years building UA implementations — custom dimensions, goals, segments, custom reports — could not simply "migrate" to GA4. They had to redesign their measurement framework from scratch on a new data model. Many organisations that delayed the migration ended up with a gap in analytics data during the transition. The lesson: platform migrations in analytics are rebuilds, not upgrades.
What fundamentally changed in GA4
| Concept | Universal Analytics | GA4 |
|---|---|---|
| Data model | Hit types (pageview, event, transaction) | All events with parameters |
| Session definition | 30-min inactivity timeout, resets at midnight | Event-based, does not reset at midnight |
| User identification | Client ID (cookie) + optional User ID | FPID + User ID + modelled signals |
| Bounce rate | Single-page sessions with no interaction | Replaced by Engagement Rate |
| Goals / Conversions | Up to 20 goals per view | Unlimited conversions (events) |
| Custom data | Custom dimensions (hit/session/user scope) | Custom dimensions (event/user scope) |
| Views | Multiple views per property | No views — data streams instead |
| Sampling | Applied in standard reports above thresholds | Reduced in standard reports, applied in Explore |
What comes next
GA4 is still relatively young — it became the primary Google Analytics product in 2023 and continues to evolve rapidly. Several directions are clear from Google's public roadmap and behaviour:
Deeper Google Ads integration. GA4's architecture is designed to feed Google's advertising ecosystem more effectively than UA was. Audiences, conversions, and signals flow more directly from GA4 into Google Ads — and this integration will deepen as Google's ad products continue to rely on first-party data.
More AI-powered analysis. Google has been adding machine learning features to GA4 — predictive audiences, anomaly detection, insights. These will become more prominent as the platform matures and Google's AI infrastructure improves.
Greater emphasis on privacy-preserving measurement. As cookie deprecation continues and privacy regulations expand, GA4's modelled data and consent mode will become more central to how the platform works. The era of complete, deterministic analytics data is ending — GA4 is Google's answer to what comes next.
Understanding where GA4 came from — two decades of evolution from a server log parser in San Diego to a privacy-first, event-based, cross-platform analytics system — makes it easier to understand where it is going and why it works the way it does today.