Google Analytics 4 is Google's event-based analytics platform that replaced Universal Analytics in July 2023. Built around privacy-first tracking and cross-platform user journeys, GA4 demands a fundamentally different implementation and reporting mindset than its predecessor.
Universal Analytics organized data around sessions and pageviews. A user arrived, viewed pages, left. GA4 abandons this structure entirely. Every interaction is an event: page_view, scroll, click, video_start, purchase. Sessions still exist as a calculated dimension, but the atomic unit is now the event with parameters attached. This matters because you can track the same user across your website and mobile app in one property, with events flowing into a single data stream. The event model also means you define what matters. Universal Analytics automatically tracked pageviews and a handful of interactions. GA4 tracks page_view and a few automatically collected events like scroll and outbound clicks, but meaningful business events require manual configuration. If you want to measure PDF downloads, form submissions, or specific button clicks, you tag those as custom events with parameters like file_name or form_id. This front-loaded configuration effort is why many teams find GA4 harder to implement. You must decide your measurement framework before data becomes useful. The upside is flexibility. Once events are flowing, you can retroactively analyze parameter combinations without redeploying tags, something impossible in Universal Analytics.
GA4 launched in a regulatory environment where third-party cookies face elimination and Apple's App Tracking Transparency curtails mobile identifiers. Google built GA4 to function in that reality. The platform uses consent mode to adjust data collection based on user permissions, modeling aggregate behavior when tracking is declined. It also employs machine learning to fill gaps: predictive metrics like purchase probability and churn probability estimate future user actions even when individual-level tracking is incomplete. Conversion modeling infers conversions that cannot be directly observed due to tracking limitations. For Canadian businesses, this becomes relevant when dealing with Quebec's Law 25 or federal PIPEDA requirements. GA4's consent mode can respect granular privacy choices while still providing directional analytics. The tradeoff is loss of precision. You no longer get clean, deterministic reporting. Some percentage of your traffic becomes modeled rather than observed. This affects attribution, audience sizes, and conversion counts. If you operate in a strict consent environment, expect GA4 numbers to feel softer than Universal Analytics once did. Machine learning also means the platform needs volume. Low-traffic sites see less accurate predictions and more data suppression due to thresholding, which hides data when user counts fall below privacy minimums.
Google stopped processing Universal Analytics data on July 1, 2023 for standard properties. Historical UA data remains accessible in read-only mode until late 2024 or early 2025, but no new hits are recorded. GA4 does not import this historical data. Your years of trend reports, conversion benchmarks, and seasonal baselines do not transfer. This creates a hard break in reporting continuity. Many organizations ran dual tracking for months before the cutoff, building parallel GA4 history while UA still processed. If you did not, you face a choice: export critical UA data to BigQuery or Data Studio before access disappears, or accept that historical context is lost. The migration is mandatory because Universal Analytics relied on cookies and client IDs in ways incompatible with current privacy regulations. Google chose to build a new platform rather than retrofit the old one. For teams managing multiple properties—franchises, multi-brand portfolios, agencies running client accounts—the migration workload multiplies. Each property requires fresh event planning, tag deployment, conversion mapping, and testing. There is no bulk conversion tool that reliably replicates UA logic in GA4. This is why agencies offering GA4 services often price migration as a distinct project rather than a configuration task.
Out of the box, GA4 captures page views, session starts, first visits, and a handful of enhanced measurement events like scrolls and site searches if you enable them. That baseline tells you almost nothing about business performance. To make GA4 actionable, you define conversions by marking specific events as conversion events in the admin interface. A purchase event becomes a conversion. A lead_form_submit event becomes a conversion. But those events must exist first, which means tagging them via Google Tag Manager or gtag.js. Unlike Universal Analytics goals, which you could configure retroactively on historical data, GA4 conversions only count forward from the moment you mark an event as a conversion. If you tag a signup event today but do not mark it as a conversion until next month, you lose that month of conversion data. This forward-only logic forces upfront planning. You also configure parameters within events to add dimensions. A purchase event might include parameters for currency, transaction_id, item count, and category. These parameters let you build custom reports and audiences. Missing a key parameter at launch means either redeploying tags or living without that dimension. Testing becomes critical. Use GA4's DebugView to confirm events and parameters fire correctly before pushing to production. Many teams discover discrepancies weeks after launch—double-counted conversions, missing parameters, events firing on wrong triggers—costing them clean baseline data.
GA4's standard reports are minimalist compared to Universal Analytics. You get basic exploration templates—funnel, path, cohort, user lifetime—but pre-built dashboards are sparse. Google expects you to build custom explorations or export data for external analysis. For small teams, this is frustrating. For advanced users, it is liberating because GA4 offers free BigQuery export on all properties, not just the 360 tier. Exporting raw event data to BigQuery lets you run SQL queries, join GA4 data with CRM or ad platform data, and build dashboards in Looker Studio or Tableau without sampling limits. This is where GA4's event model pays off. Every event and parameter lands in BigQuery as a structured record, queryable at any granularity. You can answer questions like which event sequence precedes high-value conversions or how parameter X correlates with churn without waiting for Google to build that report. The tradeoff is technical skill. If your team does not write SQL or lacks data engineering support, BigQuery remains inaccessible. Even with technical resources, schema complexity is real. GA4's nested and repeated fields require familiarity with BigQuery's array functions. Many agencies now offer GA4 services that include BigQuery setup, schema documentation, and query templates as part of the package, recognizing that the platform's value is gated behind this layer.
GA4 excels when you need cross-platform user tracking, privacy-compliant measurement, and predictive insights based on machine learning. If your business has both a website and an app, GA4 unifies those streams in ways Universal Analytics never could. If you operate in jurisdictions with strict consent laws, GA4's consent mode and modeling help you maintain some measurement when tracking is restricted. If you have enough volume and clean event data, predictive metrics can inform retention and monetization strategies. GA4 struggles when traffic is low, event tagging is inconsistent, or your team lacks analytics maturity. Thresholding hides data on small audiences. Modeled conversions introduce uncertainty. Sparse standard reports mean you cannot rely on out-of-the-box dashboards. If your primary use case was checking a few key metrics in a simple dashboard, GA4 feels like a downgrade. Universal Analytics served that need better. For agencies, this creates a service opportunity. Many clients need help not just migrating but redesigning their measurement strategy around events and conversions. The platform rewards thoughtful planning and penalizes hasty deployments. If you approach GA4 as a pure replacement for UA, you will be disappointed. If you treat it as a new measurement framework that requires rethinking what you track and why, it becomes a capable tool in a privacy-restricted environment.
Yes, but only in read-only mode and for a limited time. Google has not committed to a permanent archive. You can view historical UA reports but cannot process new data. To preserve trends and benchmarks, export critical reports to spreadsheets or connect UA to BigQuery before access is fully deprecated. Many teams export year-over-year data, conversion funnels, and top landing pages as reference baselines for comparing against GA4 going forward.
Not strictly, but highly recommended. You can deploy GA4 using the gtag.js snippet directly on your site, but custom event tracking and parameter passing become cumbersome in code. Google Tag Manager lets you configure events, triggers, and parameters through a visual interface, version changes, and test in preview mode before publishing. For any site beyond basic page view tracking, Tag Manager drastically simplifies ongoing maintenance and reduces developer dependency.
GA4 only counts conversions from the moment you designate an event as a conversion going forward. If you tag a form_submit event but wait two months to mark it as a conversion, those first two months of submissions are not retroactively counted. The event data exists in your reports and BigQuery, so you can analyze it, but it will not appear in conversion metrics or attribution reports for that period. Plan your conversion events at launch to avoid losing baseline data.
When consent mode is implemented, GA4 adjusts behavior based on user permissions. If analytics cookies are declined, GA4 can still send pings without identifiers, and Google uses conversion modeling to estimate aggregate behavior. You lose individual user tracking but retain directional metrics. Without consent mode, declining cookies typically means no data collection at all. Consent mode allows partial measurement in privacy-restricted contexts, though accuracy decreases as the proportion of non-consenting users rises.
Yes, GA4 offers free daily export to BigQuery for all properties, not just Analytics 360. Google covers export costs, but you pay standard BigQuery storage and query costs on your own GCP account. For most small to mid-sized sites, storage costs are negligible—often a few dollars per month. Query costs depend on how much data you scan. Partitioning tables by date and selecting only necessary fields keeps costs low. The free export removes a major barrier that existed in Universal Analytics, where BigQuery was a 360-only feature.
Yes, but it requires systematic event and conversion naming conventions across properties. Unlike Universal Analytics, where goals were property-specific and simple, GA4's event model means inconsistent naming makes cross-client reporting messy. Agencies often develop event taxonomy standards—common event names, required parameters, conversion definitions—and apply them at onboarding. Google Analytics 4 also supports organizations and property groups for managing access, but the lack of rollup reporting at the free tier means aggregating data across clients typically happens in BigQuery or external dashboards.