Google Analytics

Google Analytics is Google’s web and app analytics platform - the system most websites use to measure traffic, behaviour, and conversions. The current version (GA4, launched 2020 and forced as the default in 2023) replaced Universal Analytics with a fundamentally different event-based data model.

The default analytics tool for most sites because it’s free, ubiquitous, and integrates with Google Ads and Search Console. Often criticised for being harder to use than its predecessor and for relying on sampling at higher data volumes - but it’s still the baseline measurement most teams rely on.

What GA4 actually measures

Traffic acquisition. Where visitors come from - organic search, paid, direct, referral, social, email. The starting point for any channel-performance question.

User behaviour. Pageviews, events, session duration, scroll depth, page-to-page flow. The on-site activity layer.

Conversions. Pre-defined or custom events that signal business outcomes - purchases, signups, downloads, video plays, form submissions.

Audience characteristics. Demographics, interests, devices, geography. Aggregated, sampled where data is sparse.

Attribution. Multi-touch attribution across channels (last-click, data-driven, position-based, etc.). The model determines which channels get credit for which conversions.

What’s actually different about GA4 versus the old Universal Analytics

Three substantial changes:

Everything is an event. Pageviews, clicks, scrolls, conversions - all events with parameters. Reports rebuilt around event analysis rather than session-based summaries.

Cross-platform tracking. Web and app data flow into the same property. A user who visits the website then opens the app gets tracked as one user across both surfaces.

Modeled data and reduced cookie reliance. Heavy use of statistical modeling to fill gaps from privacy-related data loss (consent rejections, browser blocking, iOS limits). Useful but means GA4 numbers are less directly comparable to raw events.

Where GA4 measurement goes wrong

Three repeating patterns:

Default events not configured for the actual business model. “Conversions” out of the box are pageviews of /thank-you pages or scroll-depth thresholds. Useful for nothing. Real configuration takes 2-4 hours and most teams skip it.

Sampling on high-volume queries. Above certain query thresholds, GA4 samples data. Reports look precise but are extrapolations. Big sites need either GA4 360 (paid) or BigQuery integration to access unsampled data.

Cross-domain misconfiguration. A site with subdomains (blog.example.com → app.example.com) without proper cross-domain config sees the same user counted as two separate sessions. Easy to miss; meaningful impact on attribution accuracy.

An example

A B2B SaaS marketing team relied on GA4 for channel attribution and weekly reporting. Their numbers showed Direct traffic as the largest converter - about 40% of paid trial signups attributed to “(direct) / none.”

The audit found that “Direct” was almost entirely misclassified - visitors clicking branded ads, links from email, and links from third-party sites that were stripping the referrer. Real direct (people typing the URL) was probably under 10%.

The fix involved three things: UTM tagging discipline across all marketing surfaces, server-side measurement via BigQuery integration to recover some of the lost referrer data, and an attribution sanity check by comparing GA4 numbers to the platform-native reporting (Google Ads, LinkedIn, etc.) for cross-validation.

Three months in: “Direct” attribution dropped to about 18%. Channel allocation decisions changed accordingly - paid Search and content were under-credited and got proportionally more budget. Same actual customer journey; the data finally matched it.

We built Penfriend to produce content whose performance is clearly readable in Google Analytics. Each piece has distinct structure, distinct URLs, distinct content fingerprint - which is how content programmes get accountable to actual traffic and conversion data.

Related terms

• Google Ads - the paid platform that integrates most directly with Google Analytics
• Analytics - the broader category Google Analytics is the dominant instance of
• Conversion Rate - the metric Google Analytics is most often used to track
• Bounce Rate - a behaviour metric Google Analytics historically defined
• Customer Journey - the path Google Analytics attempts to map across sessions