Why I stopped using Google Analytics on my personal site

For years, I treated analytics as a default, the way you might leave a telly on in the background. Every new site I built shipped with a tracking snippet before it shipped with a logo, and I checked the dashboard most mornings with the same absent-mindedness I bring to the weather. Over time, the way I think about visitor data has shifted, especially after watching regulators in places like Australia treat third-party tracking with growing suspicion, and after spending more of my own hours in cafés in Melbourne and Sydney where a quiet flat white and a clean log file feel like part of the same slow pleasure.

The question stopped being whether I could collect the data and became whether I should. This is a short account of the path that took me from a dashboard obsessive to someone who runs a small personal site on a server in Dublin with no analytics script at all, and what I do instead when curiosity strikes. None of it is rocket science. It is mostly the slow rediscovery of older ways of paying attention.

The privacy side of the trade

Every page view from a stranger arrives bundled with an IP address, a user agent string, a referrer, and a fingerprint assembled from a dozen tiny details the visitor never consented to share. For a personal blog that exists mostly to publish the occasional essay, that felt like a heavy price to pay for knowing which post about Irish history got a few extra reads in March.

Australia amended its Privacy Act and introduced the Notifiable Data Breaches scheme, and the Australian Competition and Consumer Commission has spent years pushing the case that large platforms owe users a fair go rather than a clever contract. Even a humble analytics setup that loads scripts from a third party can fall under that scrutiny once a site scales or starts accepting advertising, and I would rather not be the small operator who learns the hard way what an enforceable undertaking looks like in the Federal Court.

I am not a lawyer, and I do not pretend this blog sits in the same risk bracket as a SaaS dashboard serving paying customers. But the ethical shape of the question is the same. When I cannot explain in plain English what a script does, why it is on a page, and who sees the result, my instinct is now to remove it. Privacy, like hospitality, is partly a matter of what you do not ask of a guest.

Numbers that flatter and mislead

Google Analytics has a particular talent for showing the data you want to see. Want to feel good? Look at the little upward arrow beside sessions. Want to feel anxious? Open the bounce rate panel on a post you laboured over for a week. The numbers are real, but the stories we tell from them are not, and the gap between the two has widened as dashboards have grown more elaborate.

In a small pond like mine, with maybe a few thousand readers a month scattered between Dublin, Sydney, Melbourne, and the odd visitor in São Paulo, any spike or dip is a sample of one. A post I thought was a sure thing will go unread for a fortnight. A throwaway note will pull in a hundred curious clicks from a single subreddit. The dashboards cannot tell me why, and they trained me to chase the why through metrics that were never going to answer it.

The honest data on this site, the kind I can act on, lives in the writing itself. Did I finish the piece? Did it say what I meant? Would I want to read it again in five years? That is the only review I trust now, and it does not require a JavaScript tag. The discipline of writing for a reader I will never meet has turned out to be more useful than any segment report.

The performance and weight of a tracker

A modern analytics script is not a small thing. It loads early, blocks the main thread, fires off a chain of requests, and rearranges the page experience around its own priorities. On a 4G connection on a train between Sydney and Newcastle, that delay is the difference between a clean read and a frustrating one. On a flaky café Wi-Fi in Brisbane it is worse, and a slow page is a page the reader does not come back to.

Page weight matters more than it used to, partly because Google itself now rewards Core Web Vitals and partly because readers have learned to close a tab the moment it feels heavy. I want the words to be the heaviest thing on the page, with maybe a single image and a few lines of CSS. Adding a tracker felt like bringing a megaphone into a quiet room and then trying to write a letter in the same space.

When I removed the snippet, my largest contentful paint improved, my server logs got smaller, and the site felt lighter in a way that is hard to measure but easy to feel. I had not realised how much ambient anxiety the presence of a tracker carried until it was gone, the way you do not notice a clock ticking until you take it off the wall.

Learning to read server logs

People talk about server logs as if they were a punishment, and the raw format does not help their case. Lines of text, broken requests, the occasional malformed URL, and a long parade of bots from places both friendly and otherwise. Spend a few hours with them, though, and a different picture of your site emerges. You see the bots, the humans, the real referrers, the broken links from other writers who liked something you wrote three years ago.

I set up a simple awk pipeline that pulls the things I actually care about: unique IPs per day, the top entry pages, the top exit pages, and the search terms people used to find posts through search engines. None of it ties back to a profile, and none of it is sold to anyone. It is the kind of data a librarian in Carlton or a bookseller in Newtown would feel comfortable with, and it has the virtue of being impossible to weaponise.

I also rely on the search box on the site itself. If a reader types a phrase and the search returns nothing, that is a more honest signal of a content gap than any dashboard. You can find the rest of my approach to running this kind of small site, including the writing I have done about it, in the blog section of the site.

What I do when I really want to know

There are moments when a question will not go away. A reader emails to say they could not find a particular essay, or I wonder whether a long piece about Lisbon is still pulling in traffic two years after I wrote it. In those cases I open the server logs and read them like a diary, or I ask directly. I keep an email address on the contact page and I answer everything that arrives. The signal from a person who took the time to write is worth more than a thousand pageviews, and it tends to arrive with a question I had not thought to ask myself.

For pieces that need a wider check, I still share them with a small group of friends and writers, the way you might test a recipe on a household before serving it at a dinner. If the piece survives their eyes, it is probably good enough for the rest. This is a slower loop than analytics, and that is the point. It rewards patience over impulse, and it makes the writing feel like a small act of correspondence rather than a broadcast.

When I travel, I also keep notes in a plain text file. A few days in Lisbon, for instance, gave me enough material for my travel notes on the site, and the writing informs the site as much as any metric ever did. The site has become a small archive of my own attention, and attention is the only thing I want to be measuring now.

A quiet site is not a dead site. Mine gets a steady trickle of readers, the kind who bookmark it and come back, the kind who email to correct a typo or add a thought. The numbers I do have, drawn from server logs and direct conversation, are enough. If you are weighing whether your own small project needs a tracker, the answer is probably no. A notebook, a search box, and the patience to let a post find its readers will take you further than any dashboard ever could.