Web Analytics: What It Is, Platforms, and Mobile Tracking

Web analytics is the practice of collecting and analyzing data about how people find and use your website, so you can make product and marketing decisions from evidence instead of opinion. It tells you who arrives, where they come from, what they do, and where they leave.

For a product manager, that last part is the useful one. A visitor who reads one page and leaves tells you something different from a visitor who clicks through pricing, starts a trial, and stalls on step three of sign-up. Web analytics gives you the trail.

Most teams already have a tool installed. The trouble starts when someone asks a simple question in a planning meeting, like "Did the new landing page work?", and three people open three dashboards and give three different answers. Usually the tool is fine. The team never agreed on which numbers mean what, or who looks at them.

This article covers the metrics worth agreeing on, how to think about web analytics platforms without getting lost in feature lists, and what changes when your traffic is mostly on phones.

‍

What Is Web Analytics? The Metrics That Matter

If you are asking what web analytics is in practical terms, think of four questions it answers: how many people came, how they got there, what they did, and whether they did the thing you wanted. Every metric in your dashboard maps to one of those.

A short list covers most decisions:

• Sessions and users: how much traffic you have, and how many distinct people it represents
• Traffic sources: organic search, paid, email, social, direct, and referrals
• Engagement rate: the share of visits where people actually did something
• Conversions: sign-ups, demo requests, purchases, or whatever counts as success for that page

Engagement rate deserves a closer look because it is where teams get burned. Google Analytics 4 treats a session as engaged if it lasts longer than 10 seconds, includes a key event, or has at least two page or screen views. According to Google's own documentation on engagement rate and bounce rate, bounce rate is simply the opposite: the percentage of sessions that were not engaged.

That sounds like a footnote until someone pulls an old report. Picture a PM comparing this quarter's bounce rate against a number from the previous version of Google Analytics, panicking that it dropped or spiked, and opening a ticket for the engineering team. The two tools define a bounce differently, so the comparison was never valid. Half a day lost to a definition.

Two habits prevent most of this. First, write down your definitions in one place and pin them next to the dashboard. Second, pair every traffic number with an outcome number. Ten thousand sessions mean nothing if the conversion rate sits at 0.2 percent, and a small traffic spike means a lot if conversions rose with it.

Web analytics also has a clear limit. It shows you what happened, not why. If sign-ups drop 30 percent after a redesign, the data points to the page, but only user interviews or session recordings will tell you the new button looks like an ad. Treat analytics as the smoke detector, and talk to users to find the fire. If you are earlier in that process, our guide to the product discovery process shows where this kind of data fits.

‍

Web Analytics Platforms: How to Choose One

There are dozens of web analytics platforms, and the feature comparison pages all look alike. A better way to choose is to sort them by the job you need done.

The first row is where most marketing sites start, because it is free to begin with and everyone on the team has seen the interface before. The third row matters once your question shifts from "where do visitors come from?" to "what do logged-in users do in week two?" Those are different questions, and one tool rarely answers both well. Plenty of teams run a general-purpose tool on the marketing site and a product analytics tool behind the login.

Before you compare features, answer these four questions:

1. Who will use the data every week, and what decision will they make with it?
2. Do you need to follow people across the marketing site and the product, or only measure the site?
3. Where does your data have to live, and do you need consent for tracking?
4. Who owns setup and maintenance when the person who installed it leaves?

The third question is where legal teams show up late. In the EU, regulators have taken a close look at analytics cookies. France's data protection authority, the CNIL, explains in its guidance on using analytics on websites and applications that audience measurement can be exempt from consent only under strict conditions, such as producing anonymous statistics for the publisher's own use. Whether your setup qualifies is a question for your data protection officer or legal counsel, not for the analytics vendor's sales page. Raise it before you pick a tool, not after you have six months of data you may not be allowed to keep.

The fourth question sounds boring and decides more outcomes than the rest. An analytics setup with no owner decays. Tags break after a site migration, events get renamed, and a year later nobody trusts the numbers. Name one person responsible, even if it is a part-time duty. The same discipline applies to product-led growth metrics, where unowned numbers drift fastest.

‍

Mobile Web Analytics: What Changes on Small Screens

Mobile web analytics means measuring how people behave on your website when they open it in a phone browser. It is not the same as app analytics, which uses an SDK built into a native iOS or Android app. If you have both a site and an app, you need to measure both, and you need a way to connect them if you want a single view of a customer.

The first thing that changes is what you should expect to see. Phone sessions tend to be shorter and more interrupted. People check a page while waiting for a train, get a notification, and never return. A lower time on page for mobile visitors is often normal. Judge mobile on outcomes, such as completed forms or taps on the primary button, rather than on time spent.

The second change is speed. A page that loads quickly on your office Wi-Fi may crawl on a mid-range phone over a weak connection. Google's guidance on measuring Web Vitals draws a line between lab data, collected in a controlled test, and field data, collected from real visitors. It recommends setting up your own real-user monitoring, because public datasets like the Chrome User Experience Report do not include sites with too little traffic and only show a rolling average. The target it suggests is meeting the recommended thresholds for at least 75 percent of page visits. That is a useful bar for a mobile report: segment by device, then check whether three out of four phone visitors get a fast, stable page.

Here is a simple way to run that check:

• Split your main funnel by device category and compare conversion rate on each
• Look at the step where mobile drop-off is highest, often a form or checkout field
• Test that step on a real phone, on cellular data, not in a desktop browser window resized to look small

The third change is layout. A table that reads well on a laptop can force horizontal scrolling on a phone, and a cookie banner can cover half the screen. Analytics will show you a drop-off but not the banner, so open your own site on a phone every few weeks. It takes two minutes and catches problems no dashboard will.

If mobile is a small share of your traffic, resist the urge to ignore it. Check who is in that group first. Sometimes the segment is small because the experience is poor, which makes the numbers a symptom rather than a fact about your audience.

‍