What Is Web Analytics?

Web analytics is the systematic collection, measurement and interpretation of what visitors do on a website: how many arrive, where they came from, which pages they view, what they click, and whether they end up completing a purchase or a form. It's about replacing gut feeling with decisions backed by data anyone can check.

The discipline doesn't depend on any single tool. Google Analytics 4 is the most widely used free option today, but the concept predates that platform and works the same way with any other measurement system. What changes between tools is how the data gets collected, not what actually gets measured underneath: sessions, users, traffic source and conversion events.

Applied well, web analytics ties every step of the customer journey to a concrete number: which content keeps a visitor around, where exactly the checkout process loses people, and which channel brings in the audience that actually converts. Without that data, any claim about user behavior stays a guess; with it, that claim can be checked against real numbers.

Before reading any report, it helps to know exactly what each metric counts, since some names have changed meaning between tools and versions.

Confusing GA4's bounce rate with the old Universal Analytics one is one of the most common misreadings of a report: same name, different definition.

Every visit gets logged through a measurement tag, usually a snippet of JavaScript or a tracking pixel, that fires when the page loads and sends the event to the analytics tool's server. That request should only fire after the user has given consent, since it's usually paired with a cookie that stores an identifier so the same visitor can be recognized across sessions.

Google Analytics 4 also changed the underlying data model compared to its predecessor: instead of organizing everything around the page view, it treats every interaction, from a click to a scroll, as its own event. The page view still exists, but it's just one event type among many, which lets metrics like the engaged session get calculated without depending on any single action. GA4 captures some of these events automatically once the tag is installed, things like page_view, scroll or file_download, while others, like a click on a specific button, need to be set up manually as a custom event.

A site collects these events through its own first-party cookie whenever it can, rather than relying on a third-party cookie that Safari and Firefox block by default. That's one of the reasons server-side tracking has grown: the site's own server forwards the event to the analytics platform through an API, regardless of whether the browser lets the request through.

Without web analytics, decisions about a website rest on whoever argues loudest in the meeting. With it, a pricing change, a redesigned form, or the performance of a campaign aimed at a specific target audience can be measured in actual visits, sessions and conversions, not in subjective impressions.

Specific functions sit inside this broader practice: demographic data and Google Signals add age, gender or location to sessions already logged; segmentation groups those sessions by shared traits so they can be compared against each other. Linking that data to a CRM connects what happens before the purchase, on the site, with what happens after, in the customer relationship.

Google Analytics 4 dominates the market for being free and tied into the rest of the Google ecosystem, but it isn't the only option: tools like Matomo offer a self-hosted alternative, built for teams that don't want their data sitting with an outside vendor.

• Define what counts as a conversion before implementation, not after looking at the first reports.
• Check that no tag fires before the user accepts the consent banner.
• Use segmentation to compare groups of users against each other, instead of only looking at the aggregate total.
• Connect web analytics data with the CRM to see what happens to a visitor after converting, not just before.
• Check regularly, in debug mode, that key events are still arriving correctly.
• Explain to the team that GA4's bounce rate isn't the same metric as Universal Analytics' bounce rate, before comparing numbers across periods.

• Comparing GA4's bounce rate with the old Universal Analytics one as if they were the same metric.
• Treating "total users" as real individual people without checking for duplication across a visitor's different devices.
• Turning on Google Signals without reviewing the site's consent settings first.
• Drawing conclusions from a single week of data without first ruling out bot traffic or a one-off anomaly.
• Relying only on standard browser-based tracking without checking how it holds up against Safari or Firefox traffic.