DELISORIA
Auf Deutsch
Portfolio01Magazine02Contact03 Auf Deutsch DE
Back to all articles

Magazine · · 2 min read

Counting visitors without Google Analytics

How many people read this page? A fair question — the usual answer sells out the visitors to get it. There is another way: counting in totals, on your own server.

What is wrong with the usual answer

Whoever wants to know if a website gets read usually embeds a ready-made analytics tool — and answers, in passing, questions nobody asked: which visitor returns, which paths they take, which interests can be derived. The owner wanted a number; what gets delivered is an apparatus of surveillance, complete with consent banner, processing agreements and data flowing to an advertising company.

Yet the original question is modest. How often was a page opened? Where did visits come from? Do people stay, or leave at once? None of these require knowing who anyone is.

Totals instead of profiles

Our counting stores a handful of daily totals: views per address, sessions begun, the domain of the referring site — never the full address, because what someone searched for is none of our business —, device class, language, hour. Plus dwell time and scroll depth in coarse bands: 'between 30 and 60 seconds' is a total; a per-visit number of seconds would be an individual trait. That difference is the entire point.

No IP address is stored, no identifier issued that recognises a visitor across sessions. No person can be reconstructed from a day's file — it contains only numbers. And anyone sending 'Do Not Track' is not counted at all.

The server is part of the answer

The second difference is not what, but where: the numbers are produced on our own server and stay there. No service provider reads along, no processing agreement, no third-country transfer. The counter is a single PHP endpoint, the storage a directory outside the web root, the dashboard a page that answers only to a password.

That is less convenient than an embed snippet. It is also less powerful — deliberately. Tools that can do more than their purpose requires end up being used for more than their purpose requires.

Is that enough?

For running a content site: yes. You see what gets read and what does not, where visits come from, whether a release had any effect, whether the site works on phones. What you do not see: individual people. That is no loss. A website is not an interrogation — knowing that people read is enough to write better.

A website is not an interrogation — knowing that people read is enough to write better.