What web analytics is

What web analytics is

Web analytics is the measurement of how people use a website: how many pages are opened, by how many people, where those people arrive from and what they do once they are there. The same field is also called website analytics, and in ordinary use the two names mean the same thing. Something on the site records what happens, a tool stores those records, and reports turn them into counts you can compare.

Four kinds of number carry most of it. A pageview is one load of one page. A visit, also called a session, groups the actions of one browser within one stretch of time. A visitor is the person a tool believes those visits belong to. An event is anything else you choose to record, such as a signup, a download or a click on an outgoing link. Each comes with dimensions: which page, which country, which device, which source.

For a site owner the value is in comparison rather than in the totals: which pages hold attention, which sources bring people who sign up, whether anything moved after a change. The counts carry no reason: a tool can show that a page loses people without saying why.

How the data gets collected

There are three ways to get the records, and they see different things.

A script in the page is the common one. You add a small file to your pages; it sends a request to the analytics server when a page loads, and another each time you record an event. Running in the browser, it reads what only the browser knows: the page address, the referrer, the language, the screen size, and route changes inside an application that load no new page. A request blocked by an extension or by a browser protection ends like a script that never loads: the visit happened and no record of it exists.

Events sent from your own server take the opposite route: your backend calls an endpoint at the analytics service and reports what only it knows, such as a payment that cleared or a phone order. Google documents this path for Google Analytics 4 as the Measurement Protocol and states that it augments collection from the page rather than replacing it, and that server events alone may leave you with partial reporting. Nothing here depends on the visitor’s browser, so nothing is lost to blocking, and nothing is recorded unless your code sends it.

Server log files came first and still work. A web server writes down every request it answers: the address, the time, the response code, the user agent and the referrer, and Matomo publishes an importer that loads such logs into its reports. A log holds only what the request carried: Matomo’s documentation lists screen resolution, browser plugins and page titles as missing compared with its JavaScript client, and notes that tracking cookies cannot be used this way. Logs also miss what never reaches the server, such as a page served from a cache or a content delivery network.

Put two of these on one site and the totals will not match, because the collection points, the bot lists and the rules that start a new visit all differ. A number from one tool is weak evidence for a number from another; what holds is a trend inside one tool.

The words you will meet

• A pageview is one load of one page: five pages opened by one person are five pageviews.
• A visit, or session, is one stretch of activity by one browser. Where it ends is a vendor’s setting: Google Analytics 4 and Plausible close it after 30 minutes without activity, and Matomo uses the same 30 minutes but also starts a new visit after midnight in the site’s time zone and on a new campaign source.
• A unique visitor is a tool’s estimate of how many people are behind the visits. Plausible counts a person once within a day and twice across two days; Matomo treats someone as returning from their second visit, while its cookies remain in the browser.
• A bounce is a visit that ended without what the tool treats as engagement. Plausible counts one unless the visitor opens a second page or triggers a custom event; Google Analytics 4 counts a session as engaged if it runs longer than 10 seconds, has a key event or has two or more page views, and reports the rest as bounces.
• An event is a record you send for something other than a page load, such as a form sent or a file downloaded. Check whether events count towards the volume your plan is priced on, since vendors differ.
• A goal is the action you decided matters, and a conversion is one occurrence of it; Google Analytics 4 calls these key events.
• A referrer is the address the browser reports as the origin of a click. Visits without one are grouped as direct, which mixes typed addresses with links opened from email and messaging applications.
• UTM tags are parameters you add to your own links, such as utm_source, utm_medium and utm_campaign, so a visit can be attributed to a campaign. They travel inside the address, where anyone can copy or edit them.

What web analytics does not tell you

A report counts what survived collection, and several things do not survive.

Automated traffic is the first. Crawlers, monitors and scrapers request pages all day, and each tool decides which to drop. Google Analytics 4 removes known bots using Google’s own research and the International Spiders and Bots List maintained by the Interactive Advertising Bureau, and its documentation states that you can neither switch that exclusion off nor see how much was excluded. Other tools keep their own lists, and each tool card has a row for bot and spam filtering.

Blocked and declined visits are the second. Extensions and browser protections keep some requests from leaving the page, and where a tool runs only after a consent banner is accepted, everyone who declines is missing too. Google Analytics 4 documents behavioral modelling for consent mode, which estimates the behaviour of visitors who decline analytics cookies from that of similar visitors who accept. A traffic total is a floor, not a census.

Definitions are the third: each rule in the list above is a vendor’s decision, so the same traffic yields different visit counts, bounce rates and visitor numbers in two tools at once.

Even one tool’s own numbers can be estimates: Google Analytics 4 samples when a query exceeds the property quota of 10 million events for a standard property, and marks the share of data used.

Money is absent until you put it there: a tool reports revenue only when something sends an amount with the event. The reason is absent for good, because analytics records what happened, not why.

Web analytics and privacy

Whatever the method, a request reaches the tool’s server with the page address, the referrer, the user agent and the visitor’s IP address. Many tools derive an approximate location from the IP address and then discard it, while others store it. What you send yourself travels too, such as event properties taken from a signed-in profile.

The difference between approaches is what stays in the browser. A cookie, or another identifier in browser storage, lets a tool recognise the same browser until that storage is cleared. Tools that store nothing recognise a returning visitor another way or not at all: several build a hash from the IP address and browser details with a secret value the vendor replaces on a schedule, so recognition ends when that value changes. The page on what cookieless means here goes through the mechanics.

Consent is a separate question from storage. The European Data Protection Board’s guidelines on the technical scope of Article 5(3) of the ePrivacy Directive state that the article does not apply exclusively to cookies, but also to similar technologies. Data protection law applies separately to personal data, and an IP address can be personal data. Whether your site needs a banner depends on the rules where you and your visitors are, on everything else your pages load and on how you configure the tool. Where a vendor states that no banner is needed, tool cards show that as the vendor’s position, and this page is not legal advice. The list of privacy-friendly tools sets out the conditions it applies.

How to choose a tool

Start from the requirement you cannot negotiate, because it removes most of the list in one step: where the data may be stored, whether anything may be written to the visitor’s browser, or whether you have to run the software yourself. Compare price at your traffic, hosting and script size among what is left. The table of tools holds those columns with the date each value was checked, the methodology describes the checks, and the market log records closures and changes of owner.

Questions and answers

What is the difference between web analytics and website analytics?

In practice there is none: both names describe measuring how people use a site, and vendors use them interchangeably. Web analytics is the older name of the field; website analytics reads more naturally about one particular site. Traffic estimates for sites you do not own are a different kind of service: those figures are modelled from outside a site rather than collected on it.

How is web analytics different from product analytics?

Web analytics counts what happens on pages: visits, sources and the actions you record. Product analytics follows identified users through an application over time, which is how retention, cohorts and feature adoption are measured, and usually needs a signed-in user. The line is not sharp, since several tools do some of both.

Does a small site need web analytics?

Only if a decision depends on it. A page that will not change whatever the numbers say does not need measurement. If you publish, sell or advertise, counts of pages and sources answer questions you would otherwise guess at, and the setup is one snippet or a plugin. Start with pageviews, sources and one or two events that matter; sites on WordPress can start from the tools with a WordPress plugin.

How much does it cost to start?

Nothing, for several of the options. 22 of the 44 tools in the table have a free plan in the vendor’s cloud, 16 publish their source code, and 17 can be installed on your own server, where the cost moves to the machine. Free plans differ in the volume they cover, in the unit it is counted in and in how long data is kept.

Is it legal to run web analytics in the EU?

Measuring your own site is ordinary practice, and the rules govern how you do it rather than whether you may. Two sets matter: consent rules for storing or reading information on a visitor’s device, and data protection law for personal data, which can include an IP address. The answer for one site depends on the tool, its configuration, everything else the pages load and national rules. This directory does not give legal advice; the page for tools that store data in the EU covers storage location.